diff --git a/_notebooks/2015-01-30-the-nips-experiment-examining-the-repeatability-of-peer-review.ipynb b/_notebooks/2015-01-30-the-nips-experiment-examining-the-repeatability-of-peer-review.ipynb index 62ed9f77..0e11e7d8 100644 --- a/_notebooks/2015-01-30-the-nips-experiment-examining-the-repeatability-of-peer-review.ipynb +++ b/_notebooks/2015-01-30-the-nips-experiment-examining-the-repeatability-of-peer-review.ipynb @@ -68,7 +68,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In 2014 the NeurIPS conference had 1474 active reviewers (up from 1133\n", "in 2013), 92 area chairs (up from 67 in 2013) and two program chairs,\n", @@ -110,7 +110,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Chairing a conference starts with recruitment of the program committee,\n", "which is usually done in a few stages. The primary task is to recruit\n", @@ -163,7 +163,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "With the help of [Nicolo Fusi](http://nicolofusi.com/), [Charles\n", "Twardy](http://blog.scicast.org/tag/charles-twardy/) and the entire\n", @@ -190,7 +190,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The final results of the experiment were as follows. From 170 papers 4\n", "had to be withdrawn or were rejected without completing the review\n", @@ -250,7 +250,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "There seems to have been a lot of discussion of the result, both at the\n", "conference and on bulletin boards since. Such discussion is to be\n", @@ -295,7 +295,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The first context we can place around the numbers is what would have\n", "happened at the ‘Random Conference’ where we simply accept a quarter of\n", @@ -765,7 +765,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Under the simple model we have outlined, we can be confident that there\n", "is inconsistency between two independent committees, but the level of\n", @@ -809,7 +809,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As with any decision making process there are two types of errors we can\n", "make, a type I error is accepting a paper that should be rejected. A\n", diff --git a/_notebooks/2015-09-21-peer-review-and-the-nips-experiment.ipynb b/_notebooks/2015-09-21-peer-review-and-the-nips-experiment.ipynb index 04c6a0fb..1f350e6c 100644 --- a/_notebooks/2015-09-21-peer-review-and-the-nips-experiment.ipynb +++ b/_notebooks/2015-09-21-peer-review-and-the-nips-experiment.ipynb @@ -68,7 +68,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In 2014 the NeurIPS conference had 1474 active reviewers (up from 1133\n", "in 2013), 92 area chairs (up from 67) and two program chairs, myself and\n", @@ -159,7 +159,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "With the help of [Nicolo Fusi](http://nicolofusi.com/), [Charles\n", "Twardy](http://blog.scicast.org/tag/charles-twardy/) and the entire\n", @@ -186,7 +186,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The final results of the experiment were as follows. From 170 papers 4\n", "had to be withdrawn or were rejected without completing the review\n", @@ -298,7 +298,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "There seems to have been a lot of discussion of the result, both at the\n", "conference and on bulletin boards since. Such discussion is to be\n", @@ -343,7 +343,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The first context we can place around the numbers is what would have\n", "happened at the ‘Random Conference’ where we simply accept a quarter of\n", @@ -811,7 +811,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Under the simple model we have outlined, we can be confident that there\n", "is inconsistency between two independent committees, but the level of\n", @@ -857,7 +857,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As with any decision making process there are two types of errors we can\n", "make, a type I error is accepting a paper that should be rejected. A\n", diff --git a/_notebooks/2015-10-06-matrix-factorization.ipynb b/_notebooks/2015-10-06-matrix-factorization.ipynb index 8e978a47..f2c664ef 100755 --- a/_notebooks/2015-10-06-matrix-factorization.ipynb +++ b/_notebooks/2015-10-06-matrix-factorization.ipynb @@ -352,7 +352,7 @@ "you to do is to install some software we've written for sharing\n", "information across google documents.\n", "\n", - "## `pods` \\[edit\\]\n", + "## `pods` \\[edit\\]\n", "\n", "In Sheffield we created a suite of software tools for 'Open Data\n", "Science'. Open data science is an approach to sharing code, models and\n", @@ -398,7 +398,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "# Movie Body Count Example \\[edit\\]\n", + "# Movie Body Count Example \\[edit\\]\n", "\n", "There is a crisis in the movie industry, deaths are occuring on a\n", "massive scale. In every feature film the body count is tolling up. But\n", @@ -674,7 +674,7 @@ "\n", "### Write your answer to Question 1 here\n", "\n", - "# Recommender Systems \\[edit\\]\n", + "# Recommender Systems \\[edit\\]\n", "\n", "A recommender system aims to make suggestions for items (films, books,\n", "other commercial products) given what it knows about users' tastes. The\n", @@ -797,7 +797,7 @@ "try and obtain such a lay out. To do this we first need to obtain the\n", "data.\n", "\n", - "## Obtaining the Data \\[edit\\]\n", + "## Obtaining the Data \\[edit\\]\n", "\n", "We are using a functionality of the Open Data Science software library\n", "to obtain the data. This functionality involves some prewritten code\n", @@ -943,7 +943,7 @@ "\n", "`pd.melt` do?\n", "\n", - "## Measuring Similarity \\[edit\\]\n", + "## Measuring Similarity \\[edit\\]\n", "\n", "We now need a measure for determining the similarity between the item\n", "and the user: how close the user is sitting to the item in the rooom if\n", diff --git a/_notebooks/2015-10-13-linear-regression.ipynb b/_notebooks/2015-10-13-linear-regression.ipynb index 8186be56..50e9e19d 100755 --- a/_notebooks/2015-10-13-linear-regression.ipynb +++ b/_notebooks/2015-10-13-linear-regression.ipynb @@ -327,7 +327,7 @@ " stochastic gradients.\n", "- This time: explore least squares for regression.\n", "\n", - "## Regression Examples \\[edit\\]\n", + "## Regression Examples \\[edit\\]\n", "\n", "Regression involves predicting a real value, $\\dataScalar_i$, given an\n", "input vector, $\\inputVector_i$. For example, the Tecator data involves\n", @@ -445,7 +445,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "# What is Machine Learning? \\[edit\\]\n", + "# What is Machine Learning? \\[edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -493,7 +493,7 @@ "You can also check my post blog post on [What is Machine\n", "Learning?](http://inverseprobability.com/2017/07/17/what-is-machine-learning)..\n", "\n", - "# Sum of Squares Error \\[edit\\]\n", + "# Sum of Squares Error \\[edit\\]\n", "\n", "Last week we considered a cost function for minimization of the error.\n", "We considered items (films) and users and assumed that each movie\n", @@ -524,7 +524,7 @@ "mainly inspired by the thinking of [Pierre-Simon\n", "Laplace](https://en.wikipedia.org/wiki/Pierre-Simon_Laplace).\n", "\n", - "## Regression: Linear Releationship \\[edit\\]\n", + "## Regression: Linear Releationship \\[edit\\]\n", "\n", "For many their first encounter with what might be termed a machine\n", "learning method is fitting a straight line. A straight line is\n", @@ -540,7 +540,7 @@ "a yearly basis. And $c$ is the winning pace as estimated at year 0.\n", "\n", "\\defined{overdeterminedInaugural}\n", - "## Overdetermined System \\[edit\\]\n", + "## Overdetermined System \\[edit\\]\n", "\n", "The challenge with a linear model is that it has two unknowns, $m$, and\n", "$c$. Observing data allows us to write down a system of simultaneous\n", @@ -604,7 +604,7 @@ "world, and the manner in which it was incomplete is *unknown*. His idea\n", "was that such unknowns could be dealt with through probability.\n", "\n", - "### Pierre-Simon Laplace \\[edit\\]\n", + "### Pierre-Simon Laplace \\[edit\\]\n", "\n", "\n", "\n", @@ -715,7 +715,7 @@ "the real world. Laplace's idea is that we should represent that unknown\n", "corruption with a *probability distribution*.\n", "\n", - "## A Probabilistic Process \\[edit\\]\n", + "## A Probabilistic Process \\[edit\\]\n", "\n", "However, it was left to an admirer of Gauss to develop a practical\n", "probability density for that purpose. It was Carl Friederich Gauss who\n", @@ -727,7 +727,7 @@ "a probabilistic or stochastic process, to distinguish it from a\n", "deterministic process.\n", "\n", - "## The Gaussian Density \\[edit\\]\n", + "## The Gaussian Density \\[edit\\]\n", "\n", "The Gaussian density is perhaps the most commonly used probability\n", "density. It is defined by a *mean*, $\\meanScalar$, and a *variance*,\n", @@ -744,7 +744,7 @@ "${\\dataStd}^2=0.0225$. Mean shown as red line. It could represent the\n", "heights of a population of students.\n", "\n", - "## Two Important Gaussian Properties \\[edit\\]\n", + "## Two Important Gaussian Properties \\[edit\\]\n", "\n", "The Gaussian density has many important properties, but for the moment\n", "we'll review two of them.\n", @@ -1308,7 +1308,7 @@ "\\addreading{@Rogers:book11}{For fitting linear models: Section 1.1-1.2}\n", "\\addreading{@Bishop:book06}{Section 1.2.5 up to equation 1.65}\n", "\\reading\n", - "## Objective Functions and Regression \\[edit\\]" + "## Objective Functions and Regression \\[edit\\]" ] }, { diff --git a/_notebooks/2015-10-20-basis-functions.ipynb b/_notebooks/2015-10-20-basis-functions.ipynb index b30894c4..be0de20b 100755 --- a/_notebooks/2015-10-20-basis-functions.ipynb +++ b/_notebooks/2015-10-20-basis-functions.ipynb @@ -374,7 +374,7 @@ "\n", "# Basis Functions\n", "\n", - "## Basis Functions \\[edit\\]\n", + "## Basis Functions \\[edit\\]\n", "\n", "Here's the idea, instead of working directly on the original input\n", "space, $\\inputVector$, we build models in a new space,\n", @@ -524,7 +524,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Different Bases \\[edit\\]\n", + "## Different Bases \\[edit\\]\n", "\n", "Our choice of basis can be made based on what our beliefs about what is\n", "appropriate for the data. For example, the polynomial basis extends the\n", @@ -533,7 +533,7 @@ "\\basisFunc_j(\\inputScalar_i) = \\inputScalar_i^j\n", "$$ which is known as the *polynomial basis*.\n", "\n", - "## Polynomial Basis \\[edit\\]" + "## Polynomial Basis \\[edit\\]" ] }, { @@ -643,7 +643,7 @@ "Now we look at basis functions that have been used as the *activation*\n", "functions in neural network model.\n", "\n", - "## Radial Basis Functions \\[edit\\]\n", + "## Radial Basis Functions \\[edit\\]\n", "\n", "Another type of basis is sometimes known as a 'radial basis' because the\n", "effect basis functions are constructed on 'centres' and the effect of\n", @@ -742,7 +742,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Rectified Linear Units \\[edit\\]" + "## Rectified Linear Units \\[edit\\]" ] }, { @@ -842,7 +842,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Hyperbolic Tangent Basis \\[edit\\]" + "## Hyperbolic Tangent Basis \\[edit\\]" ] }, { @@ -917,7 +917,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Fourier Basis \\[edit\\]\n", + "## Fourier Basis \\[edit\\]\n", "\n", "[Joseph Fourier](https://en.wikipedia.org/wiki/Joseph_Fourier) suggested\n", "that functions could be converted to a sum of sines and cosines. A\n", @@ -1057,7 +1057,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Fitting to Data \\[edit\\]\n", + "## Fitting to Data \\[edit\\]\n", "\n", "Now we are going to consider how these basis functions can be adjusted\n", "to fit to a particular data set. We will return to the olympic marathon\n", diff --git a/_notebooks/2015-10-27-generalization.ipynb b/_notebooks/2015-10-27-generalization.ipynb index 4fe94837..7b8da2a3 100755 --- a/_notebooks/2015-10-27-generalization.ipynb +++ b/_notebooks/2015-10-27-generalization.ipynb @@ -329,7 +329,7 @@ "- Explored the different characteristics of different basis function\n", " models\n", "\n", - "## Alan Turing \\[edit\\]\n", + "## Alan Turing \\[edit\\]\n", "\n", "\n", "\n", @@ -368,7 +368,7 @@ "the concept of generalization let's introduce some formal representation\n", "of what it means to generalize in machine learning.\n", "\n", - "## Expected Loss \\[edit\\]\n", + "## Expected Loss \\[edit\\]\n", "\n", "Our objective function so far has been the negative log likelihood,\n", "which we have minimized (via the sum of squares error) to obtain our\n", @@ -428,7 +428,7 @@ "$$ which up to the constant $\\frac{1}{\\numData}$ is identical to the\n", "objective function we have been using so far.\n", "\n", - "## Estimating Risk through Validation \\[edit\\]\n", + "## Estimating Risk through Validation \\[edit\\]\n", "\n", "Unfortuantely, minimising the empirial risk only guarantees something\n", "about our performance on the training data. If we don't have enough data\n", @@ -443,7 +443,7 @@ "our model. This means that it doesn't exhibit the same bias that the\n", "empirical risk does when estimating the true risk.\n", "\n", - "## Validation \\[edit\\]\n", + "## Validation \\[edit\\]\n", "\n", "In this lab we will explore techniques for model selection that make use\n", "of validation data. Data that isn't seen by the model in the learning\n", @@ -558,7 +558,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Validation on the Olympic Marathon Data \\[edit\\]\n", + "## Validation on the Olympic Marathon Data \\[edit\\]\n", "\n", "The first thing we'll do is fit a standard linear model to the data. We\n", "recall from previous lectures and lab classes that to do this we need to\n", @@ -703,7 +703,7 @@ "\n", "Figure: Polynomial fit to olympic data with 26 basis functions.\n", "\n", - "## Hold Out Validation on Olympic Marathon Data \\[edit\\]" + "## Hold Out Validation on Olympic Marathon Data \\[edit\\]" ] }, { @@ -896,7 +896,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Leave One Out Validation \\[edit\\]" + "## Leave One Out Validation \\[edit\\]" ] }, { @@ -973,7 +973,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## $k$-fold Cross Validation \\[edit\\]" + "## $k$-fold Cross Validation \\[edit\\]" ] }, { @@ -1058,7 +1058,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Bias Variance Decomposition \\[edit\\]\n", + "## Bias Variance Decomposition \\[edit\\]\n", "\n", "Expected test error for different variations of the *training data*\n", "sampled from, $\\Pr(\\dataVector, \\dataScalar)$\n", @@ -1076,7 +1076,7 @@ " prediction. Error due to variance is error in the model due to an\n", " overly complex model.\n", "\n", - "## Bias vs Variance Error Plots \\[edit\\]\n", + "## Bias vs Variance Error Plots \\[edit\\]\n", "\n", "Helper function for sampling data from two different classes." ] diff --git a/_notebooks/2015-11-03-bayesian-regression.ipynb b/_notebooks/2015-11-03-bayesian-regression.ipynb index e3569bf5..876e41a7 100755 --- a/_notebooks/2015-11-03-bayesian-regression.ipynb +++ b/_notebooks/2015-11-03-bayesian-regression.ipynb @@ -333,7 +333,7 @@ "$$\n", "\n", "\n", - "# Underdetermined System \\[edit\\]\n", + "# Underdetermined System \\[edit\\]\n", "\n", "What about the situation where you have more parameters than data in\n", "your simultaneous equation? This is known as an *underdetermined*\n", @@ -396,7 +396,7 @@ "uncertainty. Multiple solutions are consistent with one specified\n", "point.\n", "\n", - "## A Philosophical Dispute: Probabilistic Treatment of Parameters? \\[edit\\]\n", + "## A Philosophical Dispute: Probabilistic Treatment of Parameters? \\[edit\\]\n", "\n", "From a philosophical perspective placing a probability distribution over\n", "the *parameters* is known as the *Bayesian* approach. This is because\n", @@ -542,7 +542,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## The Bayesian Approach \\[edit\\]\n", + "## The Bayesian Approach \\[edit\\]\n", "\n", "Now we will study Bayesian approaches to regression. In the Bayesian\n", "approach we define a *prior* density over our parameters, $m$ and $c$ or\n", @@ -622,7 +622,7 @@ "already have our likelihood from our earlier discussion, so our focus\n", "now turns to the prior density.\n", "\n", - "## Prior Distribution \\[edit\\]\n", + "## Prior Distribution \\[edit\\]\n", "\n", "The tradition in Bayesian inference is to place a probability density\n", "over the parameters of interest in your model. This choice is made\n", @@ -719,14 +719,14 @@ "and\n", "$\\mu = \\frac{\\tau^2}{\\dataStd^2} \\sum_{i=1}^\\numData(\\dataScalar_i-m\\inputScalar_i)$.\n", "\n", - "## The Joint Density \\[edit\\]\n", + "## The Joint Density \\[edit\\]\n", "\n", "- Really want to know the *joint* posterior density over the\n", " parameters $c$ *and* $m$.\n", "- Could now integrate out over $m$, but it's easier to consider the\n", " multivariate case.\n", "\n", - "## Two Dimensional Gaussian \\[edit\\]\n", + "## Two Dimensional Gaussian \\[edit\\]\n", "\n", "Consider the distribution of height (in meters) of an adult male human\n", "population. We will approximate the marginal density of heights as a\n", @@ -795,7 +795,7 @@ "$$To deal with this dependence we now introduce the notion of\n", "*correlation* to the multivariate Gaussian density.\n", "\n", - "## Sampling Two Dimensional Variables \\[edit\\]" + "## Sampling Two Dimensional Variables \\[edit\\]" ] }, { @@ -829,7 +829,7 @@ "Figure: Samples from *correlated* Gaussian variables that might\n", "represent heights and weights.\n", "\n", - "## Independent Gaussians \\[edit\\]\n", + "## Independent Gaussians \\[edit\\]\n", "\n", "$$\n", "p(w, h) = p(w)p(h)\n", @@ -906,7 +906,7 @@ "\\addreading{@Bishop:book06}{Multivariate Gaussians: Section 2.3 up to top of pg 85}\n", "\\addreading{@Bishop:book06}{Section 3.3 up to 159 (pg 152–159)}\n", "\\reading\n", - "## Generating from the Model \\[edit\\]\n", + "## Generating from the Model \\[edit\\]\n", "\n", "A very important aspect of probabilistic modelling is to *sample* from\n", "your model to see what type of assumptions you are making about your\n", @@ -1201,7 +1201,7 @@ "inference, we need to computer the *posterior* distribution and sample\n", "from that density.\n", "\n", - "## Computing the Posterior \\[edit\\]\n", + "## Computing the Posterior \\[edit\\]\n", "\n", "We will now attampt to compute the *posterior distribution*. In the\n", "lecture we went through the maths that allows us to compute the\n", @@ -1411,7 +1411,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", + "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", "\n", "Five fold cross validation tests the ability of the model to\n", "*interpolate*." @@ -1531,7 +1531,7 @@ "Figure: Bayesian fit with 26th degree polynomial and five fold cross\n", "validation scores.\n", "\n", - "## Sampling from the Posterior \\[edit\\]\n", + "## Sampling from the Posterior \\[edit\\]\n", "\n", "Before we were able to sample the prior values for the mean\n", "*independently* from a Gaussian using `np.random.normal` and scaling the\n", @@ -1702,7 +1702,7 @@ "$$ This is our *implicit* assumption for $\\dataVector$ given our prior\n", "assumption for $\\mappingVector$.\n", "\n", - "## Marginal Likelihood \\[edit\\]\n", + "## Marginal Likelihood \\[edit\\]\n", "\n", "- The marginal likelihood can also be computed, it has the form: $$\n", " p(\\dataVector|\\inputMatrix, \\dataStd^2, \\alpha) = \\frac{1}{(2\\pi)^\\frac{n}{2}\\left|\\kernelMatrix\\right|^\\frac{1}{2}} \\exp\\left(-\\frac{1}{2} \\dataVector^\\top \\kernelMatrix^{-1} \\dataVector\\right)\n", @@ -1712,7 +1712,7 @@ "- So it is a zero mean $\\numData$-dimensional Gaussian with covariance\n", " matrix $\\kernelMatrix$.\n", "\n", - "## Computing the Mean and Error Bars of the Functions \\[edit\\]\n", + "## Computing the Mean and Error Bars of the Functions \\[edit\\]\n", "\n", "These ideas together, now allow us to compute the mean and error bars of\n", "the predictions. The mean prediction, before corrupting by noise is\n", diff --git a/_notebooks/2015-11-24-naive-bayes.ipynb b/_notebooks/2015-11-24-naive-bayes.ipynb index f10381da..5aff7265 100755 --- a/_notebooks/2015-11-24-naive-bayes.ipynb +++ b/_notebooks/2015-11-24-naive-bayes.ipynb @@ -321,7 +321,7 @@ "\n", "\n", "\n", - "## Introduction to Classification \\[edit\\]\n", + "## Introduction to Classification \\[edit\\]\n", "\n", "Classification is perhaps the technique most closely assocated with\n", "machine learning. In the speech based agents, on-device classifiers are\n", @@ -372,7 +372,7 @@ "appropriate *class of function*, $\\mappingFunction(\\cdot)$, to use and\n", "(3) selecting the right parameters, $\\weightVector$.\n", "\n", - "## Classification Examples \\[edit\\]\n", + "## Classification Examples \\[edit\\]\n", "\n", "- Classifiying hand written digits from binary images (automatic zip\n", " code reading)\n", @@ -382,7 +382,7 @@ "- Categorization of document types (different types of news article on\n", " the internet)\n", "\n", - "## Bernoulli Distribution \\[edit\\]\n", + "## Bernoulli Distribution \\[edit\\]\n", "\n", "Our focus has been on models where the objective function is inspired by\n", "a probabilistic analysis of the problem. In particular we've argued that\n", @@ -531,7 +531,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Maximum Likelihood in the Bernoulli \\[edit\\]\n", + "## Maximum Likelihood in the Bernoulli \\[edit\\]\n", "\n", "Maximum likelihood in the Bernoulli distribution is straightforward.\n", "Let's assume we have data, $\\dataVector$ which consists of a vector of\n", @@ -602,7 +602,7 @@ "3. Posterior distribution\n", "4. Marginal likelihood\n", "\n", - "## Naive Bayes Classifiers \\[edit\\]\n", + "## Naive Bayes Classifiers \\[edit\\]\n", "\n", "*Note*: Everything we do below is possible using standard packages like\n", "`scikit-learn`, our purpose in this session is to help you understand\n", @@ -747,7 +747,7 @@ "\\errorFunction(\\pi, \\paramVector) = \\errorFunction(\\paramVector) + \\errorFunction(\\pi).\n", "$$\n", "\n", - "## Nigerian NMIS Data \\[edit\\]\n", + "## Nigerian NMIS Data \\[edit\\]\n", "\n", "First we will load in the Nigerian NMIS health data. Our aim will be to\n", "predict whether a center has maternal health delivery services given the\n", @@ -874,7 +874,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Naive Bayes NMIS \\[edit\\]\n", + "## Naive Bayes NMIS \\[edit\\]\n", "\n", "We can now specify the naive Bayes model. For the genres we want to\n", "model the data as Bernoulli distributed, and for the year and body count\n", diff --git a/_notebooks/2015-12-01-logistic-and-glm.ipynb b/_notebooks/2015-12-01-logistic-and-glm.ipynb index 4294c9a3..42273136 100755 --- a/_notebooks/2015-12-01-logistic-and-glm.ipynb +++ b/_notebooks/2015-12-01-logistic-and-glm.ipynb @@ -368,7 +368,7 @@ "$p(\\dataVector|\\inputMatrix)$.\n", "\n", "\\addreading{@Rogers:book11}{Section 5.2.2 up to pg 182}\n", - "## Logistic Regression \\[edit\\]\n", + "## Logistic Regression \\[edit\\]\n", "\n", "A logistic regression is an approach to classification which extends the\n", "linear basis function models we've already explored. Rather than\n", @@ -487,7 +487,7 @@ "$\\mappingFunction_i = \\mappingVector^\\top \\basisVector(\\inputVector_i)$\n", "we can plot the value of the inverse link function as below.\n", "\n", - "### Sigmoid Function \\[edit\\]\n", + "### Sigmoid Function \\[edit\\]\n", "\n", "\n", "\n", @@ -710,7 +710,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Batch Gradient Descent \\[edit\\]\n", + "## Batch Gradient Descent \\[edit\\]\n", "\n", "We will need to define some initial random values for our vector and\n", "then minimize the objective by descending the gradient." @@ -832,7 +832,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Going Further: Optimization \\[edit\\]\n", + "## Going Further: Optimization \\[edit\\]\n", "\n", "Other optimization techniques for generalized linear models include\n", "[Newton's method](http://en.wikipedia.org/wiki/Newton%27s_method), it\n", @@ -854,13 +854,13 @@ "data set you can try the `pods.datasets.google_trends()` for some count\n", "data.\n", "\n", - "## Poisson Distribution \\[edit\\]\n", + "## Poisson Distribution \\[edit\\]\n", "\n", "\n", "\n", "Figure: The Poisson distribution.\n", "\n", - "## Poisson Regression \\[edit\\]\n", + "## Poisson Regression \\[edit\\]\n", "\n", "## Bayesian Approaches\n", "\n", diff --git a/_notebooks/2015-12-15-gaussian-processes.ipynb b/_notebooks/2015-12-15-gaussian-processes.ipynb index aa0cd617..e95c7069 100755 --- a/_notebooks/2015-12-15-gaussian-processes.ipynb +++ b/_notebooks/2015-12-15-gaussian-processes.ipynb @@ -356,7 +356,7 @@ "akin to the naive Bayes approach, but actually is closely related to the\n", "generalized linear model approach.\n", "\n", - "## Gaussian Processes \\[edit\\]\n", + "## Gaussian Processes \\[edit\\]\n", "\n", "Models where we model the entire joint distribution of our training\n", "data, $p(\\dataVector, \\inputMatrix)$ are sometimes described as\n", @@ -387,7 +387,7 @@ "for computing the mean and covariance. For this reason they are known as\n", "mean and covariance functions.\n", "\n", - "## Prediction Across Two Points with GPs \\[edit\\]" + "## Prediction Across Two Points with GPs \\[edit\\]" ] }, { @@ -542,7 +542,7 @@ " \\kernelMatrix= \\begin{bmatrix} \\kernelMatrix_{\\mappingFunctionVector, \\mappingFunctionVector} & \\kernelMatrix_{*, \\mappingFunctionVector}\\\\ \\kernelMatrix_{\\mappingFunctionVector, *} & \\kernelMatrix_{*, *}\\end{bmatrix}\n", " $$\n", "\n", - "## Marginal Likelihood \\[edit\\]\n", + "## Marginal Likelihood \\[edit\\]\n", "\n", "To understand the Gaussian process we're going to build on our\n", "understanding of the marginal likelihood for Bayesian regression. In the\n", @@ -870,7 +870,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Non-degenerate Gaussian Processes \\[edit\\]\n", + "## Non-degenerate Gaussian Processes \\[edit\\]\n", "\n", "The process described above is degenerate. The covariance function is of\n", "rank at most $\\numHidden$ and since the theoretical amount of data could\n", @@ -1066,7 +1066,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Bayesian Inference by Rejection Sampling \\[edit\\]\n", + "## Bayesian Inference by Rejection Sampling \\[edit\\]\n", "\n", "One view of Bayesian inference is to assume we are given a mechanism for\n", "generating samples, where we assume that mechanism is representing on\n", @@ -1668,7 +1668,7 @@ "Figure: Variation in the data fit term, the capacity term and the\n", "negative log likelihood for different lengthscales.\n", "\n", - "## Exponentiated Quadratic Covariance \\[edit\\]\n", + "## Exponentiated Quadratic Covariance \\[edit\\]\n", "\n", "The exponentiated quadratic covariance, also known as the Gaussian\n", "covariance or the RBF covariance and the squared exponential. Covariance\n", @@ -1699,7 +1699,7 @@ "
\n", "Figure: The exponentiated quadratic covariance function.\n", "\n", - "## Olympic Marathon Data \\[edit\\]\n", + "## Olympic Marathon Data \\[edit\\]\n", "\n", "\n", "\n", @@ -1790,7 +1790,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Alan Turing \\[edit\\]\n", + "## Alan Turing \\[edit\\]\n", "\n", "
\n", "\n", @@ -1943,7 +1943,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Gene Expression Example \\[edit\\]\n", + "## Gene Expression Example \\[edit\\]\n", "\n", "We now consider an example in gene expression. Gene expression is the\n", "measurement of mRNA levels expressed in cells. These mRNA levels show\n", @@ -1951,7 +1951,7 @@ "use a Gaussian process to determine whether a given gene is active, or\n", "we are merely observing a noise response.\n", "\n", - "## Della Gatta Gene Data \\[edit\\]\n", + "## Della Gatta Gene Data \\[edit\\]\n", "\n", "- Given given expression levels in the form of a time series from\n", " @DellaGatta:direct08." @@ -2231,7 +2231,7 @@ "\n", - "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", + "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", "\n", "[]{style=\"text-align:right\"}\n", "\n", @@ -2361,7 +2361,7 @@ "districts to give an immediate impression of the current status of the\n", "disease across the country.\n", "\n", - "## Additive Covariance \\[edit\\]\n", + "## Additive Covariance \\[edit\\]\n", "\n", "An additive covariance function is derived from considering the result\n", "of summing two Gaussian processes together. If the first Gaussian\n", @@ -2389,7 +2389,7 @@ "Figure: An additive covariance function formed by combining two\n", "exponentiated quadratic covariance functions.\n", "\n", - "## Analysis of US Birth Rates \\[edit\\]\n", + "## Analysis of US Birth Rates \\[edit\\]\n", "\n", "\n", "\n", @@ -2415,7 +2415,7 @@ "Figure: Two different editions of Bayesian Data Analysis\n", "[@Gelman:bayesian13].\n", "\n", - "## Basis Function Covariance \\[edit\\]\n", + "## Basis Function Covariance \\[edit\\]\n", "\n", "The fixed basis function covariance just comes from the properties of a\n", "multivariate Gaussian, if we decide $$\n", @@ -2470,7 +2470,7 @@ "Figure: A covariance function based on a non-linear basis given by\n", "$\\basisVector(\\inputVector)$.\n", "\n", - "## Brownian Covariance \\[edit\\]" + "## Brownian Covariance \\[edit\\]" ] }, { @@ -2508,7 +2508,7 @@ "
\n", "Figure: Brownian motion covariance function.\n", "\n", - "## MLP Covariance \\[edit\\]" + "## MLP Covariance \\[edit\\]" ] }, { @@ -2545,7 +2545,7 @@ "derived by considering the infinite limit of a neural network with\n", "probit activation functions.\n", "\n", - "## GPSS: Gaussian Process Summer School \\[edit\\]\n", + "## GPSS: Gaussian Process Summer School \\[edit\\]\n", "\n", "::: {style=\"width:1.5cm;text-align:center\"}\n", "\\includesvgclass{../slides/diagrams/logo/gpss-logo.svg}\n", @@ -2556,7 +2556,7 @@ "material on line. Details of the school, future events and past events\n", "can be found at the website .\n", "\n", - "## GPy: A Gaussian Process Framework in Python \\[edit\\]\n", + "## GPy: A Gaussian Process Framework in Python \\[edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2018-06-04-bayesian-methods.ipynb b/_notebooks/2018-06-04-bayesian-methods.ipynb index a5813289..a8f9b057 100755 --- a/_notebooks/2018-06-04-bayesian-methods.ipynb +++ b/_notebooks/2018-06-04-bayesian-methods.ipynb @@ -314,7 +314,7 @@ "\n", "\n", "\n", - "# What is Machine Learning? \\[edit\\]\n", + "# What is Machine Learning? \\[edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -362,7 +362,7 @@ "You can also check my post blog post on [What is Machine\n", "Learning?](http://inverseprobability.com/2017/07/17/what-is-machine-learning)..\n", "\n", - "## Olympic Marathon Data \\[edit\\]\n", + "## Olympic Marathon Data \\[edit\\]\n", "\n", "\n", "\n", @@ -459,7 +459,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Regression: Linear Releationship \\[edit\\]\n", + "## Regression: Linear Releationship \\[edit\\]\n", "\n", "For many their first encounter with what might be termed a machine\n", "learning method is fitting a straight line. A straight line is\n", @@ -475,7 +475,7 @@ "a yearly basis. And $c$ is the winning pace as estimated at year 0.\n", "\n", "\\defined{overdeterminedInaugural}\n", - "## Overdetermined System \\[edit\\]\n", + "## Overdetermined System \\[edit\\]\n", "\n", "The challenge with a linear model is that it has two unknowns, $m$, and\n", "$c$. Observing data allows us to write down a system of simultaneous\n", @@ -539,7 +539,7 @@ "world, and the manner in which it was incomplete is *unknown*. His idea\n", "was that such unknowns could be dealt with through probability.\n", "\n", - "### Pierre-Simon Laplace \\[edit\\]\n", + "### Pierre-Simon Laplace \\[edit\\]\n", "\n", "\n", "\n", @@ -650,7 +650,7 @@ "the real world. Laplace's idea is that we should represent that unknown\n", "corruption with a *probability distribution*.\n", "\n", - "## A Probabilistic Process \\[edit\\]\n", + "## A Probabilistic Process \\[edit\\]\n", "\n", "However, it was left to an admirer of Gauss to develop a practical\n", "probability density for that purpose. It was Carl Friederich Gauss who\n", @@ -662,7 +662,7 @@ "a probabilistic or stochastic process, to distinguish it from a\n", "deterministic process.\n", "\n", - "## The Gaussian Density \\[edit\\]\n", + "## The Gaussian Density \\[edit\\]\n", "\n", "The Gaussian density is perhaps the most commonly used probability\n", "density. It is defined by a *mean*, $\\meanScalar$, and a *variance*,\n", @@ -679,7 +679,7 @@ "${\\dataStd}^2=0.0225$. Mean shown as red line. It could represent the\n", "heights of a population of students.\n", "\n", - "## Two Important Gaussian Properties \\[edit\\]\n", + "## Two Important Gaussian Properties \\[edit\\]\n", "\n", "The Gaussian density has many important properties, but for the moment\n", "we'll review two of them.\n", @@ -1261,7 +1261,7 @@ "- Section 1.1-1.2 of @Rogers:book11 for fitting linear models.\n", "- Section 1.2.5 of @Bishop:book06 up to equation 1.65.\n", "\n", - "## Objective Functions and Regression \\[edit\\]" + "## Objective Functions and Regression \\[edit\\]" ] }, { @@ -2493,7 +2493,7 @@ "\n", "- Section 1.3 of @Rogers:book11 for Matrix & Vector Review.\n", "\n", - "## Basis Functions \\[edit\\]\n", + "## Basis Functions \\[edit\\]\n", "\n", "Here's the idea, instead of working directly on the original input\n", "space, $\\inputVector$, we build models in a new space,\n", @@ -2640,7 +2640,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Polynomial Fits to Olympic Data \\[edit\\]" + "## Polynomial Fits to Olympic Data \\[edit\\]" ] }, { @@ -2707,7 +2707,7 @@ "Figure: Fit of a 2 degree polynomial to the olympic marathon\n", "data.\n", "\n", - "# Underdetermined System \\[edit\\]\n", + "# Underdetermined System \\[edit\\]\n", "\n", "What about the situation where you have more parameters than data in\n", "your simultaneous equation? This is known as an *underdetermined*\n", @@ -2770,7 +2770,7 @@ "uncertainty. Multiple solutions are consistent with one specified\n", "point.\n", "\n", - "## Alan Turing \\[edit\\]\n", + "## Alan Turing \\[edit\\]\n", "\n", "
\n", "\n", @@ -2809,7 +2809,7 @@ "the concept of generalization let's introduce some formal representation\n", "of what it means to generalize in machine learning.\n", "\n", - "## Prior Distribution \\[edit\\]\n", + "## Prior Distribution \\[edit\\]\n", "\n", "The tradition in Bayesian inference is to place a probability density\n", "over the parameters of interest in your model. This choice is made\n", @@ -2906,7 +2906,7 @@ "and\n", "$\\mu = \\frac{\\tau^2}{\\dataStd^2} \\sum_{i=1}^\\numData(\\dataScalar_i-m\\inputScalar_i)$.\n", "\n", - "## Two Dimensional Gaussian \\[edit\\]\n", + "## Two Dimensional Gaussian \\[edit\\]\n", "\n", "Consider the distribution of height (in meters) of an adult male human\n", "population. We will approximate the marginal density of heights as a\n", @@ -2975,7 +2975,7 @@ "$$To deal with this dependence we now introduce the notion of\n", "*correlation* to the multivariate Gaussian density.\n", "\n", - "## Sampling Two Dimensional Variables \\[edit\\]" + "## Sampling Two Dimensional Variables \\[edit\\]" ] }, { @@ -3009,7 +3009,7 @@ "Figure: Samples from *correlated* Gaussian variables that might\n", "represent heights and weights.\n", "\n", - "## Independent Gaussians \\[edit\\]\n", + "## Independent Gaussians \\[edit\\]\n", "\n", "$$\n", "p(w, h) = p(w)p(h)\n", @@ -3052,7 +3052,7 @@ "\\covarianceMatrix = \\rotationMatrix \\mathbf{D} \\rotationMatrix^\\top\n", "$$\n", "\n", - "## Generating from the Model \\[edit\\]\n", + "## Generating from the Model \\[edit\\]\n", "\n", "A very important aspect of probabilistic modelling is to *sample* from\n", "your model to see what type of assumptions you are making about your\n", @@ -3347,7 +3347,7 @@ "inference, we need to computer the *posterior* distribution and sample\n", "from that density.\n", "\n", - "## Computing the Posterior \\[edit\\]\n", + "## Computing the Posterior \\[edit\\]\n", "\n", "We will now attampt to compute the *posterior distribution*. In the\n", "lecture we went through the maths that allows us to compute the\n", @@ -3460,7 +3460,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", + "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", "\n", "Five fold cross validation tests the ability of the model to\n", "*interpolate*." @@ -3580,7 +3580,7 @@ "Figure: Bayesian fit with 26th degree polynomial and five fold cross\n", "validation scores.\n", "\n", - "## Marginal Likelihood \\[edit\\]\n", + "## Marginal Likelihood \\[edit\\]\n", "\n", "- The marginal likelihood can also be computed, it has the form: $$\n", " p(\\dataVector|\\inputMatrix, \\dataStd^2, \\alpha) = \\frac{1}{(2\\pi)^\\frac{n}{2}\\left|\\kernelMatrix\\right|^\\frac{1}{2}} \\exp\\left(-\\frac{1}{2} \\dataVector^\\top \\kernelMatrix^{-1} \\dataVector\\right)\n", diff --git a/_notebooks/2018-11-14-bayesian-methods-abuja.ipynb b/_notebooks/2018-11-14-bayesian-methods-abuja.ipynb index 6b42d849..677749fb 100755 --- a/_notebooks/2018-11-14-bayesian-methods-abuja.ipynb +++ b/_notebooks/2018-11-14-bayesian-methods-abuja.ipynb @@ -322,7 +322,7 @@ "\n", "\n", "\n", - "# What is Machine Learning? \\[edit\\]\n", + "# What is Machine Learning? \\[edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -370,7 +370,7 @@ "You can also check my post blog post on [What is Machine\n", "Learning?](http://inverseprobability.com/2017/07/17/what-is-machine-learning)..\n", "\n", - "# Nigerian NMIS Data \\[edit\\]\n", + "# Nigerian NMIS Data \\[edit\\]\n", "\n", "As an example data set we will use Nigerian NMIS Health Facility data\n", "from openAFRICA. It can be found here\n", @@ -691,7 +691,7 @@ "in a different order. The *state* of the program is always as we left it\n", "after running the previous part.\n", "\n", - "# Probabilities \\[edit\\]\n", + "# Probabilities \\[edit\\]\n", "\n", "We are now going to do some simple review of probabilities and use this\n", "review to explore some aspects of our data.\n", @@ -720,7 +720,7 @@ "P(Y=y) \\approx \\frac{n_y}{N}.\n", "$$\n", "\n", - "## Probability and the NMIS Data \\[edit\\]\n", + "## Probability and the NMIS Data \\[edit\\]\n", "\n", "Let's use the sum rule to compute the estimate the probability that a\n", "facility has more than two nurses." @@ -814,7 +814,7 @@ "
\n", "The different basic probability distributions.\n", "
\n", - "## A Pictorial Definition of Probability \\[edit\\]\n", + "## A Pictorial Definition of Probability \\[edit\\]\n", "\n", "\n", "\n", @@ -977,7 +977,7 @@ "question we have about the world. Bayes rule (via the product rule)\n", "tells us how to *invert* the probability.\n", "\n", - "## Probabilities for Extracting Information from Data \\[edit\\]\n", + "## Probabilities for Extracting Information from Data \\[edit\\]\n", "\n", "What use is all this probability in data science? Let's think about how\n", "we might use the probabilities to do some decision making. Let's look at\n", @@ -1014,7 +1014,7 @@ "\n", "### Write your answer to Question 4 here\n", "\n", - "## Probabilistic Modelling \\[edit\\]\n", + "## Probabilistic Modelling \\[edit\\]\n", "\n", "This Bayesian approach is designed to deal with uncertainty arising from\n", "fitting our prediction function to the data we have, a reduced data set.\n", @@ -1072,7 +1072,7 @@ "$$ and we have *unsupervised learning* (from where we can get deep\n", "generative models).\n", "\n", - "## Graphical Models \\[edit\\]\n", + "## Graphical Models \\[edit\\]\n", "\n", "One way of representing a joint distribution is to consider conditional\n", "dependencies between data. Conditional dependencies allow us to\n", @@ -1113,7 +1113,7 @@ "To capture the complexity in the interelationship between the data, the\n", "graph itself becomes more complex, and less interpretable.\n", "\n", - "## Introduction to Classification \\[edit\\]\n", + "## Introduction to Classification \\[edit\\]\n", "\n", "Classification is perhaps the technique most closely assocated with\n", "machine learning. In the speech based agents, on-device classifiers are\n", @@ -1164,7 +1164,7 @@ "appropriate *class of function*, $\\mappingFunction(\\cdot)$, to use and\n", "(3) selecting the right parameters, $\\weightVector$.\n", "\n", - "## Classification Examples \\[edit\\]\n", + "## Classification Examples \\[edit\\]\n", "\n", "- Classifiying hand written digits from binary images (automatic zip\n", " code reading)\n", @@ -1174,7 +1174,7 @@ "- Categorization of document types (different types of news article on\n", " the internet)\n", "\n", - "## Bernoulli Distribution \\[edit\\]\n", + "## Bernoulli Distribution \\[edit\\]\n", "\n", "Our focus has been on models where the objective function is inspired by\n", "a probabilistic analysis of the problem. In particular we've argued that\n", @@ -1323,7 +1323,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Maximum Likelihood in the Bernoulli \\[edit\\]\n", + "## Maximum Likelihood in the Bernoulli \\[edit\\]\n", "\n", "Maximum likelihood in the Bernoulli distribution is straightforward.\n", "Let's assume we have data, $\\dataVector$ which consists of a vector of\n", @@ -1394,7 +1394,7 @@ "3. Posterior distribution\n", "4. Marginal likelihood\n", "\n", - "## Naive Bayes Classifiers \\[edit\\]\n", + "## Naive Bayes Classifiers \\[edit\\]\n", "\n", "*Note*: Everything we do below is possible using standard packages like\n", "`scikit-learn`, our purpose in this session is to help you understand\n", @@ -1539,7 +1539,7 @@ "\\errorFunction(\\pi, \\paramVector) = \\errorFunction(\\paramVector) + \\errorFunction(\\pi).\n", "$$\n", "\n", - "## Nigerian NMIS Data \\[edit\\]\n", + "## Nigerian NMIS Data \\[edit\\]\n", "\n", "First we will load in the Nigerian NMIS health data. Our aim will be to\n", "predict whether a center has maternal health delivery services given the\n", @@ -1637,7 +1637,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Naive Bayes NMIS \\[edit\\]\n", + "## Naive Bayes NMIS \\[edit\\]\n", "\n", "We can now specify the naive Bayes model. For the genres we want to\n", "model the data as Bernoulli distributed, and for the year and body count\n", diff --git a/_notebooks/2018-12-10-machine-learning-and-the-physical-world.ipynb b/_notebooks/2018-12-10-machine-learning-and-the-physical-world.ipynb index 30c9067b..6afe2fc1 100755 --- a/_notebooks/2018-12-10-machine-learning-and-the-physical-world.ipynb +++ b/_notebooks/2018-12-10-machine-learning-and-the-physical-world.ipynb @@ -56,7 +56,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -73,7 +73,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -145,7 +145,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Machine learning allows us to extract knowledge from data to form a\n", "prediction.\n", @@ -183,7 +183,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The real challenge, however, is end-to-end decision making. Taking\n", "information from the environment and using it to drive decision making\n", @@ -199,7 +199,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -233,7 +233,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -322,7 +322,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -360,7 +360,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -411,7 +411,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -495,7 +495,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Machine learning aims to replicate processes through the direct use of\n", "data. When deployed in the domain of ‘artificial intelligence,’ the\n", @@ -518,7 +518,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -593,7 +593,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> Uncertainty quantification (UQ) is the science of quantitative\n", "> characterization and reduction of uncertainties in both computational\n", @@ -638,7 +638,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To illustrate the above mentioned concepts we we use the [mountain car\n", "simulator](https://github.com/openai/gym/wiki/MountainCarContinuous-v0).\n", @@ -1088,7 +1088,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In the previous section we solved the mountain car problem by directly\n", "emulating the reward but no considerations about the dynamics $$\n", @@ -1652,7 +1652,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In some scenarios we have simulators of the same environment that have\n", "different fidelities, that is that reflect with different level of\n", @@ -1799,7 +1799,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "It is time to build the multi-fidelity model for both the position and\n", "the velocity.\n", @@ -1977,7 +1977,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2165,7 +2165,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One challenge for practitioners in Gaussian processes, is flexible\n", "software that allows the construction of the relevant GP modle. With\n", @@ -2291,7 +2291,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Common reinforcement learning methods suffer from data inefficiency,\n", "which can be a issue in real world applications where gathering\n", diff --git a/_notebooks/2019-01-09-gaussian-processes.ipynb b/_notebooks/2019-01-09-gaussian-processes.ipynb index f2e5d6c9..8846f66e 100755 --- a/_notebooks/2019-01-09-gaussian-processes.ipynb +++ b/_notebooks/2019-01-09-gaussian-processes.ipynb @@ -328,7 +328,7 @@ "Gaussian process models. It is [available freely\n", "online](http://www.gaussianprocess.org/gpml/).\n", "\n", - "# What is Machine Learning? [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "# What is Machine Learning? [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -387,7 +387,7 @@ "function, but for we will save those for another day. For the moment,\n", "let us focus on the prediction function.\n", "\n", - "## Neural Networks and Prediction Functions [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Neural Networks and Prediction Functions [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "Neural networks are adaptive non-linear function models. Originally,\n", "they were studied (by McCulloch and Pitts [@McCulloch:neuron43]) as\n", @@ -445,7 +445,7 @@ "them away and understand the family of functions that the model\n", "describes.\n", "\n", - "## Probabilistic Modelling [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Probabilistic Modelling [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "This Bayesian approach is designed to deal with uncertainty arising from\n", "fitting our prediction function to the data we have, a reduced data set.\n", @@ -503,7 +503,7 @@ "$$ and we have *unsupervised learning* (from where we can get deep\n", "generative models).\n", "\n", - "## Graphical Models [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Graphical Models [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "One way of representing a joint distribution is to consider conditional\n", "dependencies between data. Conditional dependencies allow us to\n", @@ -658,7 +658,7 @@ "$$ so the elements of the covariance or *kernel* matrix are formed by\n", "inner products of the rows of the *design matrix*.\n", "\n", - "## Gaussian Process [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Gaussian Process [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "This is the essence of a Gaussian process. Instead of making assumptions\n", "about our density over each data point, $\\dataScalar_i$ as i.i.d. we\n", @@ -677,7 +677,7 @@ "Viewing a neural network in this way is also what allows us to beform\n", "sensible *batch* normalizations [@Ioffe:batch15].\n", "\n", - "## Non-degenerate Gaussian Processes [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Non-degenerate Gaussian Processes [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "The process described above is degenerate. The covariance function is of\n", "rank at most $\\numHidden$ and since the theoretical amount of data could\n", @@ -745,7 +745,7 @@ "Radford and David were also pioneers in making their software widely\n", "available and publishing material on the web.\n", "\n", - "## Bayesian Inference by Rejection Sampling [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Bayesian Inference by Rejection Sampling [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "One view of Bayesian inference is to assume we are given a mechanism for\n", "generating samples, where we assume that mechanism is representing on\n", @@ -824,7 +824,7 @@ "\n", "\n", "\n", - "## Sampling a Function [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Sampling a Function [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "We will consider a Gaussian distribution with a particular structure of\n", "covariance matrix. We will generate *one* sample from a 25-dimensional\n", @@ -920,7 +920,7 @@ "along with the conditional distribution of $\\mappingFunction_2$ given\n", "$\\mappingFunction_1$*\n", "\n", - "## Uluru [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Uluru [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", @@ -1031,7 +1031,7 @@ " $$\\covarianceMatrix_* = \\kernelMatrix_{*,*} - \\kernelMatrix_{*,\\mappingFunctionVector}\n", " \\kernelMatrix^{-1} \\kernelMatrix_{\\mappingFunctionVector, *}$$\n", "\n", - "## Exponentiated Quadratic Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Exponentiated Quadratic Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "The exponentiated quadratic covariance, also known as the Gaussian\n", "covariance or the RBF covariance and the squared exponential. Covariance\n", @@ -1068,7 +1068,7 @@ "
\n", "*The exponentiated quadratic covariance function.*\n", "
\n", - "## Olympic Marathon Data [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Olympic Marathon Data [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", @@ -1172,7 +1172,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Alan Turing [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Alan Turing [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", @@ -1440,7 +1440,7 @@ "*Variation in the data fit term, the capacity term and the negative log\n", "likelihood for different lengthscales.*\n", "\n", - "## Della Gatta Gene Data [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Della Gatta Gene Data [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "- Given given expression levels in the form of a time series from\n", " @DellaGatta:direct08.\n", @@ -1702,7 +1702,7 @@ "\n", - "## Example: Prediction of Malaria Incidence in Uganda [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Example: Prediction of Malaria Incidence in Uganda [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "[]{style=\"text-align:right\"}\n", "\n", @@ -1854,7 +1854,7 @@ "districts to give an immediate impression of the current status of the\n", "disease across the country.\n", "\n", - "## Additive Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Additive Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "An additive covariance function is derived from considering the result\n", "of summing two Gaussian processes together. If the first Gaussian\n", @@ -1888,7 +1888,7 @@ "*An additive covariance function formed by combining two exponentiated\n", "quadratic covariance functions.*\n", "\n", - "## Analysis of US Birth Rates [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Analysis of US Birth Rates [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", @@ -1906,7 +1906,7 @@ "functions are used to take account of weekly and yearly trends. The\n", "analysis is summarized on the cover of the book.\n", "\n", - "## Basis Function Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Basis Function Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "The fixed basis function covariance just comes from the properties of a\n", "multivariate Gaussian, if we decide $$\n", @@ -1967,7 +1967,7 @@ "*A covariance function based on a non-linear basis given by\n", "$\\basisVector(\\inputVector)$.*\n", "\n", - "## Brownian Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}" + "## Brownian Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}" ] }, { @@ -2008,7 +2008,7 @@ "\n", "
\n", "
\n", - "## MLP Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}" + "## MLP Covariance [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}" ] }, { @@ -2051,7 +2051,7 @@ "considering the infinite limit of a neural network with probit\n", "activation functions.*\n", "\n", - "## GPSS: Gaussian Process Summer School [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## GPSS: Gaussian Process Summer School [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", @@ -2064,7 +2064,7 @@ "material on line. Details of the school, future events and past events\n", "can be found at the website .\n", "\n", - "## GPy: A Gaussian Process Framework in Python [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## GPy: A Gaussian Process Framework in Python [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", @@ -2086,7 +2086,7 @@ "likelihoods, multivariate outputs, dimensionality reduction and\n", "approximations for larger data sets.\n", "\n", - "## Other Software [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Other Software [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "GPy has inspired other software solutions, first of all\n", "[GPflow](https://github.com/GPflow/GPflow), which uses Tensor Flow's\n", @@ -2098,7 +2098,7 @@ "GPy itself is being restructured with MXFusion as its computational\n", "engine to give similiar capabilities.\n", "\n", - "## MXFusion: Modular Probabilistic Programming on MXNet [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## MXFusion: Modular Probabilistic Programming on MXNet [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "
\n", "\n", diff --git a/_notebooks/2019-01-09-gaussian-processes.slides.ipynb b/_notebooks/2019-01-09-gaussian-processes.slides.ipynb index 76f45682..5116674a 100755 --- a/_notebooks/2019-01-09-gaussian-processes.slides.ipynb +++ b/_notebooks/2019-01-09-gaussian-processes.slides.ipynb @@ -317,7 +317,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\#\n", "\n", "[@Rasmussen:book06]{style=\"text-align:right\"}\n", @@ -327,12 +327,12 @@ "\n", "
\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### What is Machine Learning?\n", "\n", @@ -384,7 +384,7 @@ " possible prediction functions.\n", "- Also uncertainties in objective, leave those for another day.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Neural Networks and Prediction Functions\n", "\n", "- adaptive non-linear function models inspired by simple neuron models\n", @@ -432,7 +432,7 @@ "\n", "- Consider the probabilistic approach.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Probabilistic Modelling\n", "\n", "- Probabilistically we want, $$\n", @@ -485,7 +485,7 @@ "\n", "- Prediction: $p(\\dataVector^*| \\dataVector)$\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Graphical Models\n", "\n", "- Represent joint distribution through *conditional dependencies*.\n", @@ -578,7 +578,7 @@ "\n", "### Multivariate Gaussian Properties\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Recall Univariate Gaussian Properties\n", "\n", ". . .\n", @@ -624,7 +624,7 @@ "2. Even the parameters *within* the process can be handled, by\n", " considering a particular limit.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Multivariate Gaussian Properties\n", "\n", "- If $$\n", @@ -727,7 +727,7 @@ "- Viewing a neural network in this way is also what allows us to\n", " beform sensible *batch* normalizations [@Ioffe:batch15].\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Non-degenerate Gaussian Processes\n", "\n", "- This process is *degenerate*.\n", @@ -787,7 +787,7 @@ "\n", "- David MacKay's PhD thesis [@MacKay:bayesian92]\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -910,7 +910,7 @@ "\n", "### Distributions over Functions\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -1050,7 +1050,7 @@ "\n", "
\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -1131,7 +1131,7 @@ " \\kernelMatrix = \\begin{bmatrix} \\kernelScalar_{1, 1} & \\kernelScalar_{1, 2}\\\\ \\kernelScalar_{2, 1} & \\kernelScalar_{2, 2}.\\end{bmatrix}\n", " $$\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -1279,7 +1279,7 @@ "\n", "
\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Exponentiated Quadratic Covariance" ] }, @@ -1354,7 +1354,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Olympic Marathon Data\n", "\n", @@ -1427,7 +1427,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Alan Turing\n", "\n", @@ -1502,7 +1502,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -2220,7 +2220,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Della Gatta Gene Data\n", "\n", @@ -2426,7 +2426,7 @@ "\n", "\n", "-->\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Example: Prediction of Malaria Incidence in Uganda\n", "\n", @@ -2502,7 +2502,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Additive Covariance" ] @@ -2597,7 +2597,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "{\\#\\#\\# Analysis of US Birth Rates \\#\\#\\#\n", "\n", @@ -2631,7 +2631,7 @@ "\n", "@Gelman:bayesian13\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Basis Function Covariance" ] @@ -2689,7 +2689,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Brownian Covariance" ] }, @@ -2740,7 +2740,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# MLP Covariance" ] }, @@ -2792,7 +2792,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### GPSS: Gaussian Process Summer School\n", "\n", @@ -2814,7 +2814,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### GPy: A Gaussian Process Framework in Python\n", "\n", @@ -2845,14 +2845,14 @@ "- Dimensionality reduction.\n", "- Approximations for large data sets.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Other Software\n", "\n", "- [GPflow](https://github.com/GPflow/GPflow)\n", "- [GPyTorch](https://github.com/cornellius-gp/gpytorch)\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### MXFusion: Modular Probabilistic Programming on MXNet\n", "\n", diff --git a/_notebooks/2019-01-11-deep-gaussian-processes.slides.ipynb b/_notebooks/2019-01-11-deep-gaussian-processes.slides.ipynb index b1950251..58d7762d 100755 --- a/_notebooks/2019-01-11-deep-gaussian-processes.slides.ipynb +++ b/_notebooks/2019-01-11-deep-gaussian-processes.slides.ipynb @@ -336,9 +336,9 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Approximations\n", "\n", @@ -372,7 +372,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Low Rank Motivation\n", "\n", @@ -394,7 +394,7 @@ "\n", "@Snelson:pseudo05,@Quinonero:unifying05,@Lawrence:larger07,@Titsias:variational09,@Thang:unifying17\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Variational Compression\n", "\n", "- Inducing variables are a compression of the real observations.\n", @@ -411,7 +411,7 @@ " parallelization @Gal:distributed14,@Dai:gpu14, @Seeger:auto17\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Nonparametric Gaussian Processes\n", "\n", @@ -572,7 +572,7 @@ "- Unfortunately computing $p(\\dataVector|\\inducingVector)$ is\n", " intractable\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Variational Bound on $p(\\dataVector |\\inducingVector)$\n", "\n", @@ -643,7 +643,7 @@ " parallelization @Gal:distributed14,@Dai:gpu14, @Seeger:auto17\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -773,7 +773,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Leads to Other Approximations ...\n", "\n", "- Let’s be explicity about storing approximate posterior of\n", @@ -884,9 +884,9 @@ "- *Deep Gaussian Processes and Variational Propagation of Uncertainty*\n", " @Damianou:thesis2015\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### \n", "\n", @@ -901,9 +901,9 @@ "MacKay: NeurIPS Tutorial 1997 “Have we thrown out the baby with the\n", "bathwater?” [Published as @MacKay:gpintroduction98]\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -938,7 +938,7 @@ "\\end{align}\n", "$$\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\\#\\#\\# Overfitting\n", "\n", "- Potential problem: if number of nodes in two adjacent layers is big,\n", @@ -974,7 +974,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -1037,10 +1037,10 @@ "- Equivalent to prior over parameters, take width of each layer to\n", " infinity.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Deep Face\n", "\n", @@ -1060,7 +1060,7 @@ "[Source: DeepFace\n", "[@Taigman:deepface14]]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Deep Learning as Pinball\n", "\n", @@ -1112,9 +1112,9 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Mathematically\n", "\n", @@ -1311,7 +1311,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Deep Gaussian Processes\n", "\n", @@ -1319,7 +1319,7 @@ " [@Bengio:deep09; @Hinton:fast06; @Salakhutdinov:quantitative08]\n", "- We use variational approach to stack GP models.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -1407,7 +1407,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Stacked GP" ] @@ -1504,13 +1504,13 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### GPy: A Gaussian Process Framework in Python\n", "\n", @@ -1541,9 +1541,9 @@ "- Dimensionality reduction.\n", "- Approximations for large data sets.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Olympic Marathon Data\n", "\n", @@ -1616,7 +1616,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Alan Turing\n", "\n", @@ -1821,9 +1821,9 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Della Gatta Gene Data\n", "\n", @@ -2121,7 +2121,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -2279,11 +2279,11 @@ "source": [ "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -2451,11 +2451,11 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}" + "[\\small{[edit]}]{style=\"text-align:right\"}" ] }, { @@ -2612,7 +2612,7 @@ "source": [ "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Motion Capture\n", "\n", @@ -2627,7 +2627,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "[Thanks to: Zhenwen Dai and Neil D.\n", "Lawrence]{style=\"text-align:right\"}" @@ -2757,7 +2757,7 @@ "\n", "\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Deep Health\n", "\n", @@ -2797,7 +2797,7 @@ "- Choice of which class of mathematical functions we use is a vital\n", " component of our *model*.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Emukit Playground\n", "\n", @@ -3073,7 +3073,7 @@ "250 observations of high fidelity simulator and 250 of the low fidelity\n", "simulator\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### Emukit\n", "\n", @@ -3096,9 +3096,9 @@ "- *Bayesian quadrature*: compute integrals of functions that are\n", " expensive to evaluate.\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", - "[\\small{[edit]}]{style=\"text-align:right\"}\n", + "[\\small{[edit]}]{style=\"text-align:right\"}\n", "\n", "### MXFusion: Modular Probabilistic Programming on MXNet\n", "\n", diff --git a/_notebooks/2019-05-23-meta-modelling-and-deploying-ml-software.ipynb b/_notebooks/2019-05-23-meta-modelling-and-deploying-ml-software.ipynb index 7841f5fe..eb0dbdf8 100755 --- a/_notebooks/2019-05-23-meta-modelling-and-deploying-ml-software.ipynb +++ b/_notebooks/2019-05-23-meta-modelling-and-deploying-ml-software.ipynb @@ -316,7 +316,7 @@ "\n", "# Introduction\n", "\n", - "## Peppercorns [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Peppercorns [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#peppercorn-siri-figure .figure-frame}" @@ -379,10 +379,10 @@ "You can also check my blog post on [\"Natural vs Artifical\n", "Intelligence\"](http://inverseprobability.com/2018/02/06/natural-and-artificial-intelligence)\n", "\n", - "# Deep Learning [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "# Deep Learning [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "\n", - "### DeepFace [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "### DeepFace [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#deep-face-figure .figure-frame}\n", @@ -409,7 +409,7 @@ "neural network includes more than 120 million parameters, where more\n", "than 95% come from the local and fully connected layers.\n", "\n", - "### Deep Learning as Pinball [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "### Deep Learning as Pinball [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#early-pinball-figure .figure-frame}\n", @@ -531,7 +531,7 @@ "them more data efficient and gives some robustness to adversarial\n", "examples.\n", "\n", - "## Containerization [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Containerization [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#container-2539942_1920-figure .figure-frame}\n", @@ -604,7 +604,7 @@ "\n", "> Solve Supply Chain, then solve everything else.\n", "\n", - "## Emulation [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Emulation [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#statistical-emulation-1-figure .figure-frame}\n", @@ -682,7 +682,7 @@ "we run a simulator, in which case which one, or should we attempt to\n", "acquire data from a real world intervention.\n", "\n", - "## Uncertainty Quantification [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Uncertainty Quantification [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "> Uncertainty quantification (UQ) is the science of quantitative\n", "> characterization and reduction of uncertainties in both computational\n", @@ -717,7 +717,7 @@ " than an emulator fitted only using data from the most accurate and\n", " expensive simulator.\n", "\n", - "## Mountain Car Simulator [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Mountain Car Simulator [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "To illustrate the above mentioned concepts we we use the [mountain car\n", "simulator](https://github.com/openai/gym/wiki/MountainCarContinuous-v0).\n", @@ -1063,7 +1063,7 @@ "function and using the EI helped as to find a linear controller that\n", "solves the problem.\n", "\n", - "## Data Efficient Emulation [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Data Efficient Emulation [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "In the previous section we solved the mountain car problem by directly\n", "emulating the reward but no considerations about the dynamics\n", @@ -1428,7 +1428,7 @@ "calls that we needed when applying Bayesian optimization directly on the\n", "simulator this is a great gain.\n", "\n", - "## Multi-Fidelity Emulation [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Multi-Fidelity Emulation [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "In some scenarios we have simulators of the same environment that have\n", "different fidelities, that is that reflect with different level of\n", @@ -1679,7 +1679,7 @@ "observations of the high fidelity simulator and 250 of the low fidelity\n", "simulator.\n", "\n", - "## Emukit Playground [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Emukit Playground [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "Emukit playground is a software toolkit for exploring the use of\n", "statistical emulation as a tool. It was built by [Adam\n", @@ -1724,7 +1724,7 @@ "\n", "You can explore Bayesian optimization of a taxi simulation.\n", "\n", - "## Emukit [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Emukit [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#emukit-software-page-figure .figure-frame}\n", @@ -1853,7 +1853,7 @@ "\\end{align}\n", "$$\n", "\n", - "## Olympic Marathon Data [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Olympic Marathon Data [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "\n", "\n", @@ -1963,7 +1963,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Alan Turing [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Alan Turing [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "::: {.figure}\n", "::: {#turing-run-times-figure .figure-frame}\n", @@ -2143,7 +2143,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Deep GP Fit [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## Deep GP Fit [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "Let's see if a deep Gaussian process can help here. We will construct a\n", "deep Gaussian process with one hidden layer (i.e. one Gaussian process\n", @@ -2402,7 +2402,7 @@ "followed by a layer that pushes inputs to the right is what gives the\n", "heteroschedastic noise.\n", "\n", - "## MXFusion: Modular Probabilistic Programming on MXNet [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", + "## MXFusion: Modular Probabilistic Programming on MXNet [\\[[edit]{.editsection style=\"\"}\\]]{.editsection-bracket style=\"\"}\n", "\n", "One challenge for practitioners in Gaussian processes, is flexible\n", "software that allows the construction of the relevant GP modle. With\n", diff --git a/_notebooks/2019-06-03-what-is-machine-learning.ipynb b/_notebooks/2019-06-03-what-is-machine-learning.ipynb index a031e405..0a447fd3 100755 --- a/_notebooks/2019-06-03-what-is-machine-learning.ipynb +++ b/_notebooks/2019-06-03-what-is-machine-learning.ipynb @@ -323,7 +323,7 @@ "\n", "# Introduction\n", "\n", - "## Data Science Africa \\[edit\\]\n", + "## Data Science Africa \\[edit\\]\n", "\n", "\n", "\n", @@ -365,7 +365,7 @@ "this\n", "Guardian Op-ed.\n", "\n", - "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", + "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", "\n", "[]{style=\"text-align:right\"}\n", "\n", @@ -513,7 +513,7 @@ "that we receive much of our information about the world around us\n", "through computers.\n", "\n", - "## Machine Learning in Supply Chain \\[edit\\]\n", + "## Machine Learning in Supply Chain \\[edit\\]\n", "\n", "\n", "\n", @@ -536,7 +536,7 @@ "High Peak Railway (now closed, 1820s) were all constructed to improve\n", "transportation access as the factory blossomed.\n", "\n", - "## Containerization \\[edit\\]\n", + "## Containerization \\[edit\\]\n", "\n", "\n", "\n", @@ -619,7 +619,7 @@ "responsible for the automated decision making in (probably) the world's\n", "largest AI.\n", "\n", - "## For Africa \\[edit\\]\n", + "## For Africa \\[edit\\]\n", "\n", "There is a large opportunity because infrastructures around automation\n", "are moving from physical infrastructure towards information\n", @@ -636,7 +636,7 @@ "comes from Kapchorwa district, in eastern Uganda, near the border with\n", "Kenya.\n", "\n", - "## Olympic Marathon Data \\[edit\\]\n", + "## Olympic Marathon Data \\[edit\\]\n", "\n", "
\n", "\n", @@ -733,7 +733,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Polynomial Fits to Olympic Data \\[edit\\]" + "## Polynomial Fits to Olympic Data \\[edit\\]" ] }, { @@ -800,7 +800,7 @@ "Figure: Fit of a 2 degree polynomial to the olympic marathon\n", "data.\n", "\n", - "## What does Machine Learning do? \\[edit\\]\n", + "## What does Machine Learning do? \\[edit\\]\n", "\n", "Any process of automation allows us to scale what we do by codifying a\n", "process in some way that makes it efficient and repeatable. Machine\n", @@ -874,7 +874,7 @@ "functions are explored more often as they tend to improve quality of\n", "predictions but at the expense of interpretability.\n", "\n", - "## What is Machine Learning? \\[edit\\]\n", + "## What is Machine Learning? \\[edit\\]\n", "\n", "Machine learning allows us to extract knowledge from data to form a\n", "prediction.\n", @@ -906,7 +906,7 @@ "information from the enviroment and using it to drive decision making to\n", "achieve goals.\n", "\n", - "## Artificial Intelligence and Data Science \\[edit\\]\n", + "## Artificial Intelligence and Data Science \\[edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -976,7 +976,7 @@ "full randomized control trial (referred to as A/B testing in modern\n", "internet parlance).\n", "\n", - "## Neural Networks and Prediction Functions \\[edit\\]\n", + "## Neural Networks and Prediction Functions \\[edit\\]\n", "\n", "Neural networks are adaptive non-linear function models. Originally,\n", "they were studied (by McCulloch and Pitts [@McCulloch:neuron43]) as\n", @@ -1036,7 +1036,7 @@ "2. Unsupervised learning\n", "3. Reinforcement learning\n", "\n", - "# Supervised Learning \\[edit\\]\n", + "# Supervised Learning \\[edit\\]\n", "\n", "Supervised learning is one of the most widely deployed machine learning\n", "technologies, and a particular domain of success has been\n", @@ -1128,7 +1128,7 @@ "product, then the function would need some periodic components to\n", "reflect seasonal or weekly effects.\n", "\n", - "## Analysis of US Birth Rates \\[edit\\]\n", + "## Analysis of US Birth Rates \\[edit\\]\n", "\n", "\n", "\n", @@ -1193,7 +1193,7 @@ "water and protein content of meat samples was predicted as a function of\n", "the absorption of infrared light.\n", "\n", - "# Deep Learning \\[edit\\]\n", + "# Deep Learning \\[edit\\]\n", "\n", "Classical statistical models and simple machine learning models have a\n", "great deal in common. The main difference between the fields is\n", @@ -1222,7 +1222,7 @@ "end (good prediction) rather than an end in themselves (interpretable).\n", "\n", "\n", - "### DeepFace \\[edit\\]\n", + "### DeepFace \\[edit\\]\n", "\n", "\n", "\n", @@ -1237,7 +1237,7 @@ "neural network includes more than 120 million parameters, where more\n", "than 95% come from the local and fully connected layers.\n", "\n", - "### Deep Learning as Pinball \\[edit\\]\n", + "### Deep Learning as Pinball \\[edit\\]\n", "\n", "\n", "\n", @@ -1399,7 +1399,7 @@ "ability. This is the system's ability to predict in areas where it\n", "hasn't previously seen data.\n", "\n", - "## Hold Out Validation on Olympic Marathon Data \\[edit\\]" + "## Hold Out Validation on Olympic Marathon Data \\[edit\\]" ] }, { @@ -1592,7 +1592,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Bias Variance Decomposition \\[edit\\]\n", + "## Bias Variance Decomposition \\[edit\\]\n", "\n", "Expected test error for different variations of the *training data*\n", "sampled from, $\\Pr(\\dataVector, \\dataScalar)$\n", @@ -1610,7 +1610,7 @@ " prediction. Error due to variance is error in the model due to an\n", " overly complex model.\n", "\n", - "## Bias vs Variance Error Plots \\[edit\\]\n", + "## Bias vs Variance Error Plots \\[edit\\]\n", "\n", "Helper function for sampling data from two different classes." ] @@ -1860,7 +1860,7 @@ "for model selection that validation error cannot be used as an unbiased\n", "estimate of the generalization performance.\n", "\n", - "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", + "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", "\n", "Five fold cross validation tests the ability of the model to\n", "*interpolate*." @@ -1981,7 +1981,7 @@ "validation scores.\n", "\n", "\n", - "# Unsupervised Learning \\[edit\\]\n", + "# Unsupervised Learning \\[edit\\]\n", "\n", "In unsupervised learning you have data, $\\inputVector$, but no labels\n", "$\\dataScalar$. The aim in unsupervised learning is to extract structure\n", @@ -2014,7 +2014,7 @@ "discrete label) and methods that represent the data as a continuous\n", "value.\n", "\n", - "## Clustering \\[edit\\]\n", + "## Clustering \\[edit\\]\n", "\n", "Clustering methods associate each data point with a different label.\n", "Unlike in classification the label is not provided by a human annotator.\n", @@ -2167,7 +2167,7 @@ "algorithms that decompose data in more complex ways, but they can\n", "normally only be applied to smaller data sets.\n", "\n", - "## Dimensionality Reduction \\[edit\\]\n", + "## Dimensionality Reduction \\[edit\\]\n", "\n", "Dimensionality reduction methods compress the data by replacing the\n", "original data with a reduced number of continuous variables. One way of\n", @@ -2335,7 +2335,7 @@ "while returning results with acceptable latency is a particular\n", "challenge.\n", "\n", - "# Reinforcement Learning \\[edit\\]\n", + "# Reinforcement Learning \\[edit\\]\n", "\n", "The final domain of learning we will review is known as reinforcement\n", "learning. The domain of reinforcement learning is one that many\n", @@ -2536,7 +2536,7 @@ "of these functions, as dictated by their parameters, is determined by\n", "acquiring data from the real world.\n", "\n", - "## Deployment \\[edit\\]\n", + "## Deployment \\[edit\\]\n", "\n", "The methods we have introduced are roughly speaking introduced in order\n", "of difficulty of deployment. While supervised learning is more involved\n", diff --git a/_notebooks/2019-06-06-the-three-ds-of-machine-learning.ipynb b/_notebooks/2019-06-06-the-three-ds-of-machine-learning.ipynb index fc69b51a..d092d3e1 100755 --- a/_notebooks/2019-06-06-the-three-ds-of-machine-learning.ipynb +++ b/_notebooks/2019-06-06-the-three-ds-of-machine-learning.ipynb @@ -327,14 +327,14 @@ "\n", "# Introduction\n", "\n", - "## The Centrifugal Governor \\[edit\\]\n", + "## The Centrifugal Governor \\[edit\\]\n", "\n", "\n", "\n", "Figure: Centrifugal governor as held by \"Science\" on Holborn\n", "Viaduct\n", "\n", - "## Boulton and Watt's Steam Engine \\[edit\\]\n", + "## Boulton and Watt's Steam Engine \\[edit\\]\n", "\n", "\n", "\n", @@ -396,7 +396,7 @@ "broken, this would be obvious to the engine operator at start up time.\n", "The machine could be repaired before operation.\n", "\n", - "## What is Machine Learning? \\[edit\\]\n", + "## What is Machine Learning? \\[edit\\]\n", "\n", "Machine learning allows us to extract knowledge from data to form a\n", "prediction.\n", @@ -424,13 +424,13 @@ "surfacing in two different but overlapping domains: data science and\n", "artificial intelligence.\n", "\n", - "## From Model to Decision \\[edit\\]\n", + "## From Model to Decision \\[edit\\]\n", "\n", "The real challenge, however, is end-to-end decision making. Taking\n", "information from the environment and using it to drive decision making\n", "to achieve goals.\n", "\n", - "## Artificial Intelligence and Data Science \\[edit\\]\n", + "## Artificial Intelligence and Data Science \\[edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -502,7 +502,7 @@ "\n", "## Amazon: Bits and Atoms\n", "\n", - "## Machine Learning in Supply Chain \\[edit\\]\n", + "## Machine Learning in Supply Chain \\[edit\\]\n", "\n", "\n", "\n", @@ -561,7 +561,7 @@ "initially constructed by the economic need for moving goods. To improve\n", "supply chain.\n", "\n", - "## Containerization \\[edit\\]\n", + "## Containerization \\[edit\\]\n", "\n", "\n", "\n", @@ -649,7 +649,7 @@ "\n", "> Solve Supply Chain, then solve everything else.\n", "\n", - "# The Three Ds of Machine Learning Systems Design \\[edit\\]\n", + "# The Three Ds of Machine Learning Systems Design \\[edit\\]\n", "\n", "We can characterize the challenges for integrating machine learning\n", "within our systems as the three Ds. Decomposition, Data and Deployment.\n", @@ -662,7 +662,7 @@ "focus on *supervised learning* because this is arguably the technology\n", "that is best understood within machine learning.\n", "\n", - "## Decomposition \\[edit\\]\n", + "## Decomposition \\[edit\\]\n", "\n", "Machine learning is not magical pixie dust, we cannot simply automate\n", "all decisions through data. We are constrained by our data (see below)\n", @@ -760,7 +760,7 @@ "tension which we should always retain in our minds when performing our\n", "system design.\n", "\n", - "## Data \\[edit\\]\n", + "## Data \\[edit\\]\n", "\n", "It is difficult to overstate the importance of data. It is half of the\n", "equation for machine learning but is often utterly neglected. We can\n", @@ -789,7 +789,7 @@ "improved programming paradigms (object orientated, functional) and\n", "better tools (CVS, then SVN, then git).\n", "\n", - "## The Data Crisis \\[edit\\]\n", + "## The Data Crisis \\[edit\\]\n", "\n", "Anecdotally, talking to data modelling scientists. Most say they spend\n", "80% of their time acquiring and cleaning data. This is precipitating\n", @@ -894,7 +894,7 @@ "Technology Readiness Levels which attempt to quantify the readiness of\n", "technologies for deployment.b\n", "\n", - "### Three Grades of Data Readiness \\[edit\\]\n", + "### Three Grades of Data Readiness \\[edit\\]\n", "\n", "Data-readiness describes, at its coarsest level, three separate stages\n", "of data graduation.\n", @@ -925,9 +925,9 @@ "because of the need for programming skills, but the day to day problems\n", "faced are very different.\n", "\n", - "## Combining Data and Systems Design \\[edit\\]\n", + "## Combining Data and Systems Design \\[edit\\]\n", "\n", - "## Data Science as Debugging \\[edit\\]\n", + "## Data Science as Debugging \\[edit\\]\n", "\n", "One challenge for existing information technology professionals is\n", "realizing the extent to which a software ecosystem based on data differs\n", @@ -1018,7 +1018,7 @@ "where applicable, but we need to develop them to become *data first*\n", "organizations. Data needs to be cleaned at *output* not at *input*.\n", "\n", - "## Deployment \\[edit\\]\n", + "## Deployment \\[edit\\]\n", "\n", "Much of the academic machine learning systems point of view is based on\n", "a software systems point of view that is around 20 years out of date. In\n", @@ -1081,7 +1081,7 @@ "*stateless*, internal state is deployed on streams alongside external\n", "state. This allows for rapid assessment of other services' data.\n", "\n", - "## Data Oriented Architectures \\[edit\\]\n", + "## Data Oriented Architectures \\[edit\\]\n", "\n", "In a streaming architecture we shift from management of services, to\n", "management of data streams. Instead of worrying about availability of\n", @@ -1103,7 +1103,7 @@ "data retention or recomputation can be taken at the systems level rather\n", "than the component level.\n", "\n", - "## Apache Flink \\[edit\\]\n", + "## Apache Flink \\[edit\\]\n", "\n", "[Apache Flink](https://en.wikipedia.org/wiki/Apache_Flink) is a stream\n", "processing framework. Flink is a foundation for event driven processing.\n", @@ -1335,7 +1335,7 @@ "Our aim is to release our first version of a data-oriented programming\n", "environment by end of June 2019 (pending internal approval).\n", "\n", - "## Conclusion \\[edit\\]\n", + "## Conclusion \\[edit\\]\n", "\n", "We operate in a technologically evolving environment. Machine learning\n", "is becoming a key coponent in our decision-making capabilities, our\n", diff --git a/_notebooks/2019-09-10-introduction-to-deep-gps.ipynb b/_notebooks/2019-09-10-introduction-to-deep-gps.ipynb index 9259884f..77f6f82a 100755 --- a/_notebooks/2019-09-10-introduction-to-deep-gps.ipynb +++ b/_notebooks/2019-09-10-introduction-to-deep-gps.ipynb @@ -373,7 +373,7 @@ "\\end{align}\n", "$$\n", "\n", - "## Overfitting \\[edit\\]\n", + "## Overfitting \\[edit\\]\n", "\n", "One potential problem is that as the number of nodes in two adjacent\n", "layers increases, the number of parameters in the affine transformation\n", @@ -413,7 +413,7 @@ "Figure: Pictorial representation of the low rank form of the matrix\n", "$\\mappingMatrix$.\n", "\n", - "## Bottleneck Layers in Deep Neural Networks \\[edit\\]" + "## Bottleneck Layers in Deep Neural Networks \\[edit\\]" ] }, { @@ -459,7 +459,7 @@ "\\end{align}\n", "$$\n", "\n", - "## Cascade of Gaussian Processes \\[edit\\]\n", + "## Cascade of Gaussian Processes \\[edit\\]\n", "\n", "Now if we replace each of these neural networks with a Gaussian process.\n", "This is equivalent to taking the limit as the width of each layer goes\n", @@ -474,10 +474,10 @@ "\\end{align}\n", "$$\n", "\n", - "# Deep Learning \\[edit\\]\n", + "# Deep Learning \\[edit\\]\n", "\n", "\n", - "### DeepFace \\[edit\\]\n", + "### DeepFace \\[edit\\]\n", "\n", "\n", "\n", @@ -492,7 +492,7 @@ "neural network includes more than 120 million parameters, where more\n", "than 95% come from the local and fully connected layers.\n", "\n", - "### Deep Learning as Pinball \\[edit\\]\n", + "### Deep Learning as Pinball \\[edit\\]\n", "\n", "\n", "\n", @@ -603,7 +603,7 @@ "Figure: More usually deep probabilistic models are written vertically\n", "rather than horizontally as in the Markov chain.\n", "\n", - "## Why Deep? \\[edit\\]\n", + "## Why Deep? \\[edit\\]\n", "\n", "If the result of composing many functions together is simply another\n", "function, then why do we bother? The key point is that we can change the\n", @@ -649,7 +649,7 @@ "conditional dependencies. Here we are adding a side note from the\n", "chain.\n", "\n", - "## Difficulty for Probabilistic Approaches \\[edit\\]\n", + "## Difficulty for Probabilistic Approaches \\[edit\\]\n", "\n", "The challenge for composition of probabilistic models is that you need\n", "to propagate a probability densities through non linear mappings. This\n", @@ -672,7 +672,7 @@ "Figure: A Gaussian density over the input of a non linear function\n", "leads to a very non Gaussian output. Here the output is multimodal.\n", "\n", - "## Standard Variational Approach Fails \\[edit\\]\n", + "## Standard Variational Approach Fails \\[edit\\]\n", "\n", "- Standard variational bound has the form: $$\n", " \\likelihoodBound = \\expDist{\\log p(\\dataVector|\\latentMatrix)}{q(\\latentMatrix)} + \\KL{q(\\latentMatrix)}{p(\\latentMatrix)}\n", @@ -766,7 +766,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Stacked PCA \\[edit\\]\n", + "## Stacked PCA \\[edit\\]\n", "\n", "\n", "\n", @@ -811,7 +811,7 @@ "\\end{bmatrix}}$$ which is a highly structured Gaussian covariance with\n", "hierarchical dependencies between the variables $\\latentMatrix_i$.\n", "\n", - "## Stacked GP \\[edit\\]" + "## Stacked GP \\[edit\\]" ] }, { @@ -864,7 +864,7 @@ "the densities described by the deep GP are more general than those\n", "mentioned in either of these papers.\n", "\n", - "## Stacked GPs (video by David Duvenaud) \\[edit\\]" + "## Stacked GPs (video by David Duvenaud) \\[edit\\]" ] }, { @@ -887,7 +887,7 @@ "David Duvenaud also created a YouTube video to help visualize what\n", "happens as you drop through the layers of a deep GP.\n", "\n", - "## GPy: A Gaussian Process Framework in Python \\[edit\\]\n", + "## GPy: A Gaussian Process Framework in Python \\[edit\\]\n", "\n", "\n", "\n", @@ -943,7 +943,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Olympic Marathon Data \\[edit\\]\n", + "## Olympic Marathon Data \\[edit\\]\n", "\n", "
\n", "\n", @@ -1040,7 +1040,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Alan Turing \\[edit\\]\n", + "## Alan Turing \\[edit\\]\n", "\n", "
\n", "\n", @@ -1193,7 +1193,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Deep GP Fit \\[edit\\]\n", + "## Deep GP Fit \\[edit\\]\n", "\n", "Let's see if a deep Gaussian process can help here. We will construct a\n", "deep Gaussian process with one hidden layer (i.e. one Gaussian process\n", @@ -1403,7 +1403,7 @@ "followed by a layer that pushes inputs to the right is what gives the\n", "heteroschedastic noise.\n", "\n", - "## Gene Expression Example \\[edit\\]\n", + "## Gene Expression Example \\[edit\\]\n", "\n", "We now consider an example in gene expression. Gene expression is the\n", "measurement of mRNA levels expressed in cells. These mRNA levels show\n", @@ -1411,7 +1411,7 @@ "use a Gaussian process to determine whether a given gene is active, or\n", "we are merely observing a noise response.\n", "\n", - "## Della Gatta Gene Data \\[edit\\]\n", + "## Della Gatta Gene Data \\[edit\\]\n", "\n", "- Given given expression levels in the form of a time series from\n", " @DellaGatta:direct08." @@ -1829,7 +1829,7 @@ "uncertainty is the cumulative uncertainty from all the layers. Pinball\n", "plot of the della Gatta gene expression data.\n", "\n", - "## Step Function \\[edit\\]\n", + "## Step Function \\[edit\\]\n", "\n", "Next we consider a simple step function data set." ] @@ -2159,7 +2159,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Motorcycle Helmet Data \\[edit\\]\n", + "## Motorcycle Helmet Data \\[edit\\]\n", "\n", "\n", "\n", @@ -2358,7 +2358,7 @@ "Figure: Pinball plot for the mapping from input to output layer for\n", "the motorcycle helmet accelerometer data.\n", "\n", - "## Motion Capture \\[edit\\]\n", + "## Motion Capture \\[edit\\]\n", "\n", "- 'High five' data.\n", "- Model learns structure between two interacting subjects.\n", @@ -2379,7 +2379,7 @@ "latent variables that are specific to each of the figures\n", "separately.\n", "\n", - "## Fitting a GP to the USPS Digits Data \\[edit\\]\n", + "## Fitting a GP to the USPS Digits Data \\[edit\\]\n", "\n", "[Thanks to: Zhenwen Dai and Neil D.\n", "Lawrence]{style=\"text-align:right\"}\n", @@ -2742,7 +2742,7 @@ "mapping the tour into the data space. We visualize the mean of the\n", "mapping in the images.\n", "\n", - "## Deep Health \\[edit\\]\n", + "## Deep Health \\[edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2019-10-15-data-quality-and-data-readiness.ipynb b/_notebooks/2019-10-15-data-quality-and-data-readiness.ipynb index 0f49ed0c..a46a4150 100755 --- a/_notebooks/2019-10-15-data-quality-and-data-readiness.ipynb +++ b/_notebooks/2019-10-15-data-quality-and-data-readiness.ipynb @@ -329,7 +329,7 @@ "- Classical computer science separates code and data.\n", "- Machine learning short-circuits this separation.\n", "\n", - "## The Data Crisis \\[edit\\]\n", + "## The Data Crisis \\[edit\\]\n", "\n", "Anecdotally, talking to data modelling scientists. Most say they spend\n", "80% of their time acquiring and cleaning data. This is precipitating\n", @@ -390,7 +390,7 @@ "area, [AI for Data\n", "Analytics](https://www.turing.ac.uk/research_projects/artificial-intelligence-data-analytics/).\n", "\n", - "## Data Science as Debugging \\[edit\\]\n", + "## Data Science as Debugging \\[edit\\]\n", "\n", "One challenge for existing information technology professionals is\n", "realizing the extent to which a software ecosystem based on data differs\n", @@ -462,9 +462,9 @@ "provide the understanding of how to data-wrangle. Companies must fill\n", "this gap.\n", "\n", - "## Data Readiness Levels \\[edit\\]\n", + "## Data Readiness Levels \\[edit\\]\n", "\n", - "### Data Readiness Levels \\[edit\\]\n", + "### Data Readiness Levels \\[edit\\]\n", "\n", "[Data Readiness\n", "Levels](http://inverseprobability.com/2017/01/12/data-readiness-levels)\n", @@ -474,7 +474,7 @@ "Technology Readiness Levels which attempt to quantify the readiness of\n", "technologies for deployment.b\n", "\n", - "### Three Grades of Data Readiness \\[edit\\]\n", + "### Three Grades of Data Readiness \\[edit\\]\n", "\n", "Data-readiness describes, at its coarsest level, three separate stages\n", "of data graduation.\n", @@ -588,7 +588,7 @@ "practice needs to be understood by a wider community before that can\n", "happen.\n", "\n", - "## Data Oriented Architectures \\[edit\\]\n", + "## Data Oriented Architectures \\[edit\\]\n", "\n", "In a streaming architecture we shift from management of services, to\n", "management of data streams. Instead of worrying about availability of\n", @@ -610,7 +610,7 @@ "data retention or recomputation can be taken at the systems level rather\n", "than the component level.\n", "\n", - "## Apache Flink \\[edit\\]\n", + "## Apache Flink \\[edit\\]\n", "\n", "[Apache Flink](https://en.wikipedia.org/wiki/Apache_Flink) is a stream\n", "processing framework. Flink is a foundation for event driven processing.\n", @@ -778,7 +778,7 @@ "\n", "This best guess may well be driven by previous data.\n", "\n", - "## Ride Sharing: Service Oriented to Data Oriented \\[edit\\]\n", + "## Ride Sharing: Service Oriented to Data Oriented \\[edit\\]\n", "\n", "\n", "\n", @@ -874,7 +874,7 @@ "\n", "This may require a quick estimate of the ride availability.\n", "\n", - "## Information Dynamics \\[edit\\]\n", + "## Information Dynamics \\[edit\\]\n", "\n", "With all the second guessing within a complex automated decision-making\n", "system, there are potential problems with information dynamics, the\n", diff --git a/_notebooks/2019-10-21-what-is-machine-learning-ashesi.ipynb b/_notebooks/2019-10-21-what-is-machine-learning-ashesi.ipynb index ef9e3173..89bafa38 100755 --- a/_notebooks/2019-10-21-what-is-machine-learning-ashesi.ipynb +++ b/_notebooks/2019-10-21-what-is-machine-learning-ashesi.ipynb @@ -323,7 +323,7 @@ "\n", "# Introduction\n", "\n", - "## Data Science Africa \\[edit\\]\n", + "## Data Science Africa \\[edit\\]\n", "\n", "\n", "\n", @@ -365,7 +365,7 @@ "this\n", "Guardian Op-ed.\n", "\n", - "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", + "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", "\n", "[]{style=\"text-align:right\"}\n", "\n", @@ -513,7 +513,7 @@ "that we receive much of our information about the world around us\n", "through computers.\n", "\n", - "## Machine Learning in Supply Chain \\[edit\\]\n", + "## Machine Learning in Supply Chain \\[edit\\]\n", "\n", "\n", "\n", @@ -572,7 +572,7 @@ "initially constructed by the economic need for moving goods. To improve\n", "supply chain.\n", "\n", - "## Containerization \\[edit\\]\n", + "## Containerization \\[edit\\]\n", "\n", "\n", "\n", @@ -639,7 +639,7 @@ "models we use, it becomes harder to develop efficient algorithms to\n", "match those models to data.\n", "\n", - "## For Africa \\[edit\\]\n", + "## For Africa \\[edit\\]\n", "\n", "There is a large opportunity because infrastructures around automation\n", "are moving from physical infrastructure towards information\n", @@ -658,7 +658,7 @@ "these parameters to change the behavior of the function. The choice of\n", "mathematical function we use is a vital component of the model.\n", "\n", - "# Nigerian NMIS Data \\[edit\\]\n", + "# Nigerian NMIS Data \\[edit\\]\n", "\n", "As an example data set we will use Nigerian NMIS Health Facility data\n", "from openAFRICA. It can be found here\n", @@ -979,7 +979,7 @@ "in a different order. The *state* of the program is always as we left it\n", "after running the previous part.\n", "\n", - "## What does Machine Learning do? \\[edit\\]\n", + "## What does Machine Learning do? \\[edit\\]\n", "\n", "Any process of automation allows us to scale what we do by codifying a\n", "process in some way that makes it efficient and repeatable. Machine\n", @@ -1053,7 +1053,7 @@ "functions are explored more often as they tend to improve quality of\n", "predictions but at the expense of interpretability.\n", "\n", - "## What is Machine Learning? \\[edit\\]\n", + "## What is Machine Learning? \\[edit\\]\n", "\n", "Machine learning allows us to extract knowledge from data to form a\n", "prediction.\n", @@ -1081,13 +1081,13 @@ "surfacing in two different but overlapping domains: data science and\n", "artificial intelligence.\n", "\n", - "## From Model to Decision \\[edit\\]\n", + "## From Model to Decision \\[edit\\]\n", "\n", "The real challenge, however, is end-to-end decision making. Taking\n", "information from the environment and using it to drive decision making\n", "to achieve goals.\n", "\n", - "## Artificial Intelligence and Data Science \\[edit\\]\n", + "## Artificial Intelligence and Data Science \\[edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -1157,7 +1157,7 @@ "full randomized control trial (referred to as A/B testing in modern\n", "internet parlance).\n", "\n", - "## Neural Networks and Prediction Functions \\[edit\\]\n", + "## Neural Networks and Prediction Functions \\[edit\\]\n", "\n", "Neural networks are adaptive non-linear function models. Originally,\n", "they were studied (by McCulloch and Pitts [@McCulloch:neuron43]) as\n", @@ -1225,7 +1225,7 @@ "machine learning. By scanning across the image we can also determine\n", "where the animal is in the image.\n", "\n", - "## Introduction to Classification \\[edit\\]\n", + "## Introduction to Classification \\[edit\\]\n", "\n", "Classification is perhaps the technique most closely assocated with\n", "machine learning. In the speech based agents, on-device classifiers are\n", @@ -1276,7 +1276,7 @@ "appropriate *class of function*, $\\mappingFunction(\\cdot)$, to use and\n", "(3) selecting the right parameters, $\\weightVector$.\n", "\n", - "## Classification Examples \\[edit\\]\n", + "## Classification Examples \\[edit\\]\n", "\n", "- Classifiying hand written digits from binary images (automatic zip\n", " code reading)\n", @@ -1291,7 +1291,7 @@ "Figure: The perceptron algorithm.\n", "\n", "\\addreading{@Rogers:book11}{Section 5.2.2 up to pg 182}\n", - "## Logistic Regression \\[edit\\]\n", + "## Logistic Regression \\[edit\\]\n", "\n", "A logistic regression is an approach to classification which extends the\n", "linear basis function models we've already explored. Rather than\n", @@ -1410,7 +1410,7 @@ "$\\mappingFunction_i = \\mappingVector^\\top \\basisVector(\\inputVector_i)$\n", "we can plot the value of the inverse link function as below.\n", "\n", - "### Sigmoid Function \\[edit\\]\n", + "### Sigmoid Function \\[edit\\]\n", "\n", "\n", "\n", @@ -1601,7 +1601,7 @@ "gradient descent* or the Robbins Munro [@Robbins:stoch51] optimization\n", "procedure worked best for function minimization.\n", "\n", - "## Nigerian NMIS Data \\[edit\\]\n", + "## Nigerian NMIS Data \\[edit\\]\n", "\n", "First we will load in the Nigerian NMIS health data. Our aim will be to\n", "predict whether a center has maternal health delivery services given the\n", @@ -1699,7 +1699,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Batch Gradient Descent \\[edit\\]\n", + "## Batch Gradient Descent \\[edit\\]\n", "\n", "We will need to define some initial random values for our vector and\n", "then minimize the objective by descending the gradient." @@ -1821,7 +1821,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Regression \\[edit\\]\n", + "## Regression \\[edit\\]\n", "\n", "Classification is the case where our prediction function gives a\n", "discrete valued output, normally associated with a 'class'. Regression\n", @@ -1834,7 +1834,7 @@ "and 'extrapolation', which is the practice of predicting a function\n", "value beyond the regime where we have data.\n", "\n", - "## Regression Examples \\[edit\\]\n", + "## Regression Examples \\[edit\\]\n", "\n", "Regression involves predicting a real value, $\\dataScalar_i$, given an\n", "input vector, $\\inputVector_i$. For example, the Tecator data involves\n", @@ -1941,7 +1941,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Polynomial Fits to Olympic Data \\[edit\\]" + "## Polynomial Fits to Olympic Data \\[edit\\]" ] }, { @@ -2008,7 +2008,7 @@ "Figure: Fit of a 2 degree polynomial to the olympic marathon\n", "data.\n", "\n", - "## Supervised Learning Challenges \\[edit\\]\n", + "## Supervised Learning Challenges \\[edit\\]\n", "\n", "There are three principal challenges in constructing a problem for\n", "supervised learning.\n", @@ -2019,7 +2019,7 @@ " $\\mappingFunction(\\cdot)$.\n", "3. selecting the right parameters, $\\weightVector$.\n", "\n", - "## Feature Selection \\[edit\\]\n", + "## Feature Selection \\[edit\\]\n", "\n", "Feature selection is a critical stage in the algorithm design process.\n", "In the Olympic prediction example above we're only using time to predict\n", @@ -2047,7 +2047,7 @@ "feature sets used. Facebook in particular has made heavy investments in\n", "machine learning pipelines for evaluation of the feature utility.\n", "\n", - "## Class of Function, $\\mappingFunction(\\cdot)$ \\[edit\\]\n", + "## Class of Function, $\\mappingFunction(\\cdot)$ \\[edit\\]\n", "\n", "By class of function we mean, what are the characteristics of the\n", "mapping between $\\mathbf{x}$ and $y$. Often, we might choose it to be a\n", @@ -2056,7 +2056,7 @@ "product, then the function would need some periodic components to\n", "reflect seasonal or weekly effects.\n", "\n", - "## Analysis of US Birth Rates \\[edit\\]\n", + "## Analysis of US Birth Rates \\[edit\\]\n", "\n", "\n", "\n", @@ -2118,7 +2118,7 @@ "and protein content of meat samples was predicted as a function of the\n", "absorption of infrared light.\n", "\n", - "## Class of Function: Neural Networks \\[edit\\]\n", + "## Class of Function: Neural Networks \\[edit\\]\n", "\n", "One class of function that has become popular recently is neural network\n", "functions, in particular deep neural networks. The ImageNet challenge\n", @@ -2130,7 +2130,7 @@ "invariances and scale invariances are also applicable for object\n", "detection in images.\n", "\n", - "# Deep Learning \\[edit\\]\n", + "# Deep Learning \\[edit\\]\n", "\n", "Classical statistical models and simple machine learning models have a\n", "great deal in common. The main difference between the fields is\n", @@ -2159,7 +2159,7 @@ "end (good prediction) rather than an end in themselves (interpretable).\n", "\n", "\n", - "### DeepFace \\[edit\\]\n", + "### DeepFace \\[edit\\]\n", "\n", "\n", "\n", @@ -2174,7 +2174,7 @@ "neural network includes more than 120 million parameters, where more\n", "than 95% come from the local and fully connected layers.\n", "\n", - "### Deep Learning as Pinball \\[edit\\]\n", + "### Deep Learning as Pinball \\[edit\\]\n", "\n", "\n", "\n", @@ -2278,7 +2278,7 @@ "machine learning algorithms focus on simpler concepts such as linearity\n", "or smoothness.\n", "\n", - "## Parameter Estimation: Objective Functions \\[edit\\]\n", + "## Parameter Estimation: Objective Functions \\[edit\\]\n", "\n", "Once we have a set of features, and the class of functions we use is\n", "determined, we need to find the parameters of the model.\n", @@ -2329,7 +2329,7 @@ "if we design a face detector using Californians may not perform well\n", "when deployed in Kampala, Uganda.\n", "\n", - "## Generalization and Overfitting \\[edit\\]\n", + "## Generalization and Overfitting \\[edit\\]\n", "\n", "Once a supervised learning system is trained it can be placed in a\n", "sequential pipeline to automate a process that used to be done manually.\n", @@ -2350,7 +2350,7 @@ "ability. This is the system's ability to predict in areas where it\n", "hasn't previously seen data.\n", "\n", - "## Hold Out Validation on Olympic Marathon Data \\[edit\\]" + "## Hold Out Validation on Olympic Marathon Data \\[edit\\]" ] }, { @@ -2543,7 +2543,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## Bias Variance Decomposition \\[edit\\]\n", + "## Bias Variance Decomposition \\[edit\\]\n", "\n", "Expected test error for different variations of the *training data*\n", "sampled from, $\\Pr(\\dataVector, \\dataScalar)$\n", @@ -2561,7 +2561,7 @@ " prediction. Error due to variance is error in the model due to an\n", " overly complex model.\n", "\n", - "## Bias vs Variance Error Plots \\[edit\\]\n", + "## Bias vs Variance Error Plots \\[edit\\]\n", "\n", "Helper function for sampling data from two different classes." ] @@ -2811,7 +2811,7 @@ "for model selection that validation error cannot be used as an unbiased\n", "estimate of the generalization performance.\n", "\n", - "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", + "## Olympic Data with Bayesian Polynomials \\[edit\\]\n", "\n", "Five fold cross validation tests the ability of the model to\n", "*interpolate*." @@ -2932,7 +2932,7 @@ "validation scores.\n", "\n", "\n", - "# Unsupervised Learning \\[edit\\]\n", + "# Unsupervised Learning \\[edit\\]\n", "\n", "In unsupervised learning you have data, $\\inputVector$, but no labels\n", "$\\dataScalar$. The aim in unsupervised learning is to extract structure\n", @@ -2965,7 +2965,7 @@ "discrete label) and methods that represent the data as a continuous\n", "value.\n", "\n", - "## Clustering \\[edit\\]\n", + "## Clustering \\[edit\\]\n", "\n", "Clustering methods associate each data point with a different label.\n", "Unlike in classification the label is not provided by a human annotator.\n", @@ -3118,7 +3118,7 @@ "algorithms that decompose data in more complex ways, but they can\n", "normally only be applied to smaller data sets.\n", "\n", - "## Dimensionality Reduction \\[edit\\]\n", + "## Dimensionality Reduction \\[edit\\]\n", "\n", "Dimensionality reduction methods compress the data by replacing the\n", "original data with a reduced number of continuous variables. One way of\n", @@ -3286,7 +3286,7 @@ "while returning results with acceptable latency is a particular\n", "challenge.\n", "\n", - "# Reinforcement Learning \\[edit\\]\n", + "# Reinforcement Learning \\[edit\\]\n", "\n", "The final domain of learning we will review is known as reinforcement\n", "learning. The domain of reinforcement learning is one that many\n", @@ -3487,7 +3487,7 @@ "of these functions, as dictated by their parameters, is determined by\n", "acquiring data from the real world.\n", "\n", - "## Deployment \\[edit\\]\n", + "## Deployment \\[edit\\]\n", "\n", "The methods we have introduced are roughly speaking introduced in order\n", "of difficulty of deployment. While supervised learning is more involved\n", diff --git a/_notebooks/2020-01-24-r250-gp-intro.ipynb b/_notebooks/2020-01-24-r250-gp-intro.ipynb index dfbe056e..c6c1a8c5 100755 --- a/_notebooks/2020-01-24-r250-gp-intro.ipynb +++ b/_notebooks/2020-01-24-r250-gp-intro.ipynb @@ -317,7 +317,7 @@ "\n", "\n", "\n", - "### Pierre-Simon Laplace \\[edit\\]\n", + "### Pierre-Simon Laplace \\[edit\\]\n", "\n", "\n", "\n", @@ -405,7 +405,7 @@ "And *probability* is the tool we use to incorporate this ignorance\n", "leading to uncertainty or *doubt* in our predictions.\n", "\n", - "## Bayesian Inference by Rejection Sampling \\[edit\\]\n", + "## Bayesian Inference by Rejection Sampling \\[edit\\]\n", "\n", "One view of Bayesian inference is to assume we are given a mechanism for\n", "generating samples, where we assume that mechanism is representing on\n", @@ -473,7 +473,7 @@ "*posterior*). The Gaussian process allows us to do this\n", "analytically.\n", "\n", - "# What is Machine Learning? \\[edit\\]\n", + "# What is Machine Learning? \\[edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -532,7 +532,7 @@ "function, but for we will save those for another day. For the moment,\n", "let us focus on the prediction function.\n", "\n", - "## Neural Networks and Prediction Functions \\[edit\\]\n", + "## Neural Networks and Prediction Functions \\[edit\\]\n", "\n", "Neural networks are adaptive non-linear function models. Originally,\n", "they were studied (by McCulloch and Pitts [@McCulloch:neuron43]) as\n", @@ -591,7 +591,7 @@ "wish to fit parameters at all, rather we wish to integrate them away and\n", "understand the family of functions that the model describes.\n", "\n", - "## Probabilistic Modelling \\[edit\\]\n", + "## Probabilistic Modelling \\[edit\\]\n", "\n", "This Bayesian approach is designed to deal with uncertainty arising from\n", "fitting our prediction function to the data we have, a reduced data set.\n", @@ -649,7 +649,7 @@ "$$ and we have *unsupervised learning* (from where we can get deep\n", "generative models).\n", "\n", - "## Graphical Models \\[edit\\]\n", + "## Graphical Models \\[edit\\]\n", "\n", "One way of representing a joint distribution is to consider conditional\n", "dependencies between data. Conditional dependencies allow us to\n", @@ -809,7 +809,7 @@ "$$ so the elements of the covariance or *kernel* matrix are formed by\n", "inner products of the rows of the *design matrix*.\n", "\n", - "## Gaussian Process \\[edit\\]\n", + "## Gaussian Process \\[edit\\]\n", "\n", "This is the essence of a Gaussian process. Instead of making assumptions\n", "about our density over each data point, $\\dataScalar_i$ as i.i.d. we\n", @@ -828,7 +828,7 @@ "Viewing a neural network in this way is also what allows us to beform\n", "sensible *batch* normalizations [@Ioffe:batch15].\n", "\n", - "## Non-degenerate Gaussian Processes \\[edit\\]\n", + "## Non-degenerate Gaussian Processes \\[edit\\]\n", "\n", "The process described above is degenerate. The covariance function is of\n", "rank at most $\\numHidden$ and since the theoretical amount of data could\n", @@ -893,7 +893,7 @@ "\n", "\n", "\n", - "## Sampling a Function \\[edit\\]\n", + "## Sampling a Function \\[edit\\]\n", "\n", "We will consider a Gaussian distribution with a particular structure of\n", "covariance matrix. We will generate *one* sample from a 25-dimensional\n", @@ -990,7 +990,7 @@ "$\\mappingFunction_2$ along with the conditional distribution of\n", "$\\mappingFunction_2$ given $\\mappingFunction_1$\n", "\n", - "## Uluru \\[edit\\]\n", + "## Uluru \\[edit\\]\n", "\n", "\n", "\n", @@ -1107,7 +1107,7 @@ " $$\\covarianceMatrix_* = \\kernelMatrix_{*,*} - \\kernelMatrix_{*,\\mappingFunctionVector}\n", " \\kernelMatrix^{-1} \\kernelMatrix_{\\mappingFunctionVector, *}$$\n", "\n", - "## Exponentiated Quadratic Covariance \\[edit\\]\n", + "## Exponentiated Quadratic Covariance \\[edit\\]\n", "\n", "The exponentiated quadratic covariance, also known as the Gaussian\n", "covariance or the RBF covariance and the squared exponential. Covariance\n", @@ -1138,7 +1138,7 @@ "
\n", "Figure: The exponentiated quadratic covariance function.\n", "\n", - "## Olympic Marathon Data \\[edit\\]\n", + "## Olympic Marathon Data \\[edit\\]\n", "\n", "\n", "\n", @@ -1235,7 +1235,7 @@ "\n", "More recent years see more consistently quick marathons.\n", "\n", - "## Alan Turing \\[edit\\]\n", + "## Alan Turing \\[edit\\]\n", "\n", "
\n", "\n", @@ -1484,7 +1484,7 @@ "Figure: Variation in the data fit term, the capacity term and the\n", "negative log likelihood for different lengthscales.\n", "\n", - "## Gene Expression Example \\[edit\\]\n", + "## Gene Expression Example \\[edit\\]\n", "\n", "We now consider an example in gene expression. Gene expression is the\n", "measurement of mRNA levels expressed in cells. These mRNA levels show\n", @@ -1492,7 +1492,7 @@ "use a Gaussian process to determine whether a given gene is active, or\n", "we are merely observing a noise response.\n", "\n", - "## Della Gatta Gene Data \\[edit\\]\n", + "## Della Gatta Gene Data \\[edit\\]\n", "\n", "- Given given expression levels in the form of a time series from\n", " @DellaGatta:direct08.\n", @@ -1766,7 +1766,7 @@ "\n", - "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", + "## Example: Prediction of Malaria Incidence in Uganda \\[edit\\]\n", "\n", "[]{style=\"text-align:right\"}\n", "\n", @@ -1896,7 +1896,7 @@ "districts to give an immediate impression of the current status of the\n", "disease across the country.\n", "\n", - "## Additive Covariance \\[edit\\]\n", + "## Additive Covariance \\[edit\\]\n", "\n", "An additive covariance function is derived from considering the result\n", "of summing two Gaussian processes together. If the first Gaussian\n", @@ -1924,7 +1924,7 @@ "Figure: An additive covariance function formed by combining two\n", "exponentiated quadratic covariance functions.\n", "\n", - "## Analysis of US Birth Rates \\[edit\\]\n", + "## Analysis of US Birth Rates \\[edit\\]\n", "\n", "\n", "\n", @@ -1950,7 +1950,7 @@ "Figure: Two different editions of Bayesian Data Analysis\n", "[@Gelman:bayesian13].\n", "\n", - "## Basis Function Covariance \\[edit\\]\n", + "## Basis Function Covariance \\[edit\\]\n", "\n", "The fixed basis function covariance just comes from the properties of a\n", "multivariate Gaussian, if we decide $$\n", @@ -2005,7 +2005,7 @@ "Figure: A covariance function based on a non-linear basis given by\n", "$\\basisVector(\\inputVector)$.\n", "\n", - "## Brownian Covariance \\[edit\\]" + "## Brownian Covariance \\[edit\\]" ] }, { @@ -2043,7 +2043,7 @@ "
\n", "Figure: Brownian motion covariance function.\n", "\n", - "## MLP Covariance \\[edit\\]" + "## MLP Covariance \\[edit\\]" ] }, { @@ -2080,7 +2080,7 @@ "derived by considering the infinite limit of a neural network with\n", "probit activation functions.\n", "\n", - "## RELU Covariance \\[edit\\]" + "## RELU Covariance \\[edit\\]" ] }, { @@ -2114,7 +2114,7 @@ "\n", "Figure: Rectified linear unit covariance function.\n", "\n", - "## Sinc Covariance \\[edit\\]\n", + "## Sinc Covariance \\[edit\\]\n", "\n", "Another approach to developing covariance function exploits Bochner's\n", "theorem @Bochner:book59. Bochner's theorem tells us that any positve\n", @@ -2153,7 +2153,7 @@ "metadata": {}, "source": [ "\\includecovariane{sinc}{\\kernelScalar(\\inputVector, \\inputVector^\\prime) = \\alpha \\text{sinc}\\left(\\pi w r\\right)}{Sinc covariance function.}\n", - "## Polynomial Covariance \\[edit\\]\n", + "## Polynomial Covariance \\[edit\\]\n", "\n", "
\n", "$$\\kernelScalar(\\inputVector, \\inputVector^\\prime) = \\alpha(w \\inputVector^\\top\\inputVector^\\prime + b)^d$$\n", @@ -2170,7 +2170,7 @@ "\n", "Figure: Polynomial covariance function.\n", "\n", - "## Periodic Covariance \\[edit\\]\n", + "## Periodic Covariance \\[edit\\]\n", "\n", "
\n", "$$\\kernelScalar(\\inputVector, \\inputVector^\\prime) = \\alpha\\exp\\left(\\frac{-2\\sin(\\pi rw)^2}{\\lengthScale^2}\\right)$$\n", @@ -2187,7 +2187,7 @@ "\n", "Figure: Periodic covariance function.\n", "\n", - "## Linear Model of Coregionalization Covariance \\[edit\\]" + "## Linear Model of Coregionalization Covariance \\[edit\\]" ] }, { @@ -2247,7 +2247,7 @@ "\n", "Figure: Linear model of coregionalization covariance function.\n", "\n", - "## Intrinsic Coregionalization Model Covariance \\[edit\\]" + "## Intrinsic Coregionalization Model Covariance \\[edit\\]" ] }, { diff --git a/_notebooks/2021-02-02-introduction-to-machine-intelligence.ipynb b/_notebooks/2021-02-02-introduction-to-machine-intelligence.ipynb index 7f37c02e..9c115e7e 100644 --- a/_notebooks/2021-02-02-introduction-to-machine-intelligence.ipynb +++ b/_notebooks/2021-02-02-introduction-to-machine-intelligence.ipynb @@ -97,7 +97,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -304,7 +304,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> For instance, the temperature at which ice melts is found to be always\n", "> the same under ordinary circumstances, though, as we shall see, it is\n", @@ -432,7 +432,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -510,7 +510,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -659,7 +659,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -707,7 +707,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -731,7 +731,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -882,7 +882,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The [Planck space\n", "craft](https://en.wikipedia.org/wiki/Planck_(spacecraft)) was a European\n", @@ -1139,7 +1139,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This example is taken from [Thomas House’s blog\n", "post](https://personalpages.manchester.ac.uk/staff/thomas.house/blog/modelling-herd-immunity.html)\n", diff --git a/_notebooks/2021-05-05-ml-and-the-physical-world-sheffield.ipynb b/_notebooks/2021-05-05-ml-and-the-physical-world-sheffield.ipynb index 07541edc..5add2965 100644 --- a/_notebooks/2021-05-05-ml-and-the-physical-world-sheffield.ipynb +++ b/_notebooks/2021-05-05-ml-and-the-physical-world-sheffield.ipynb @@ -67,7 +67,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -112,7 +112,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -124,7 +124,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -289,7 +289,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -351,7 +351,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -375,7 +375,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -392,7 +392,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -464,7 +464,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Machine learning allows us to extract knowledge from data to form a\n", "prediction.\n", @@ -502,7 +502,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The real challenge, however, is end-to-end decision making. Taking\n", "information from the environment and using it to drive decision making\n", @@ -518,7 +518,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -552,7 +552,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -641,7 +641,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -679,7 +679,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -730,7 +730,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -814,7 +814,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Machine learning aims to replicate processes through the direct use of\n", "data. When deployed in the domain of ‘artificial intelligence,’ the\n", @@ -837,7 +837,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -912,7 +912,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> Uncertainty quantification (UQ) is the science of quantitative\n", "> characterization and reduction of uncertainties in both computational\n", @@ -957,7 +957,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To illustrate the above mentioned concepts we we use the [mountain car\n", "simulator](https://github.com/openai/gym/wiki/MountainCarContinuous-v0).\n", @@ -1407,7 +1407,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In the previous section we solved the mountain car problem by directly\n", "emulating the reward but no considerations about the dynamics $$\n", @@ -1971,7 +1971,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In some scenarios we have simulators of the same environment that have\n", "different fidelities, that is that reflect with different level of\n", @@ -2118,7 +2118,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "It is time to build the multi-fidelity model for both the position and\n", "the velocity.\n", @@ -2296,7 +2296,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2484,7 +2484,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One challenge for practitioners in Gaussian processes, is flexible\n", "software that allows the construction of the relevant GP modle. With\n", @@ -2610,7 +2610,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Common reinforcement learning methods suffer from data inefficiency,\n", "which can be a issue in real world applications where gathering\n", diff --git a/_notebooks/2021-05-17-post-digital-transformation-intellectual-debt.ipynb b/_notebooks/2021-05-17-post-digital-transformation-intellectual-debt.ipynb index cdb9003b..623e5c08 100644 --- a/_notebooks/2021-05-17-post-digital-transformation-intellectual-debt.ipynb +++ b/_notebooks/2021-05-17-post-digital-transformation-intellectual-debt.ipynb @@ -94,7 +94,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -140,7 +140,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -186,7 +186,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -212,7 +212,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -223,7 +223,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -311,7 +311,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -368,7 +368,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -466,7 +466,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -565,7 +565,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -627,7 +627,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focussed on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -670,7 +670,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -747,7 +747,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> There are three types of lies: lies, damned lies and statistics\n", ">\n", @@ -906,7 +906,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -950,7 +950,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1041,7 +1041,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox is the modern phenomenon of “as we collect more\n", "data, we understand less.” It is emerging in several domains, political\n", @@ -1093,7 +1093,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox has a sister: the big model paradox. As we build\n", "more and more complex models, we start believing that we have a\n", @@ -1111,7 +1111,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -1189,7 +1189,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -1204,7 +1204,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -1287,7 +1287,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -1573,7 +1573,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1671,7 +1671,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1682,7 +1682,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The DELVE Initiative was convened by the Royal Society early in the\n", "pandemic in response for a perceived need to increase provide policy\n", @@ -1715,7 +1715,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "- First contact *3rd April*\n", "- First meeting *7th April*\n", @@ -1816,7 +1816,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1827,7 +1827,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Corporate culture turns out to be an important component of how you can\n", "react to digital transformation. Amazon is a company that likes to take\n", @@ -1920,7 +1920,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Any policy question can be framed in a number of different ways - what\n", "are the health outcomes; what is the impact on NHS capacity; how are\n", @@ -1938,7 +1938,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To improve communication, we need to ‘externalise cognition’: have\n", "objects that are outside our brains, are persistent in the real world,\n", diff --git a/_notebooks/2021-06-16-the-neurips-experiment.ipynb b/_notebooks/2021-06-16-the-neurips-experiment.ipynb index 33bdd97d..b3184f0b 100644 --- a/_notebooks/2021-06-16-the-neurips-experiment.ipynb +++ b/_notebooks/2021-06-16-the-neurips-experiment.ipynb @@ -77,7 +77,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In 2014 the NeurIPS conference had 1474 active reviewers (up from 1133\n", "in 2013), 92 area chairs (up from 67 in 2013) and two program chairs,\n", @@ -119,7 +119,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Chairing a conference starts with recruitment of the program committee,\n", "which is usually done in a few stages. The primary task is to recruit\n", @@ -172,7 +172,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The instructions to reviewers for the 2014 conference are still\n", "available [online\n", @@ -358,7 +358,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "With the help of [Nicolo Fusi](http://nicolofusi.com/), [Charles\n", "Twardy](http://blog.scicast.org/tag/charles-twardy/) and the entire\n", @@ -385,7 +385,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The results of the experiment were as follows. From 170 papers 4 had to\n", "be withdrawn or were rejected without completing the review process, for\n", @@ -511,7 +511,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "There seems to have been a lot of discussion of the result, both at the\n", "conference and on bulletin boards since. Such discussion is to be\n", @@ -556,7 +556,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The first context we can place around the numbers is what would have\n", "happened at the ‘Random Conference’ where we simply accept a quarter of\n", @@ -1033,7 +1033,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Calibration of reviewers is the process where different interpretations\n", "of the reviewing scale are addressed. The tradition of calibration goes\n", @@ -1049,7 +1049,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In this note book we deal with reviewer calibration. Our assumption is\n", "that the score from the $j$th reviwer for the $i$th paper is given by $$\n", @@ -1129,7 +1129,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1380,7 +1380,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now we wish to predict the bias corrected scores for the papers. That\n", "involves considering a variable $s_{i,j} = f_i + e_{i,j}$ which is the\n", @@ -1411,7 +1411,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can now sample from this posterior distribution of bias-adjusted\n", "scores jointly, to get a set of scores for all papers. For this set of\n", @@ -1523,7 +1523,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Here is the histogram of the reviewer scores after calibration." ] @@ -1742,7 +1742,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "For NeurIPS 2014 we experimented with duplicate papers: we pushed papers\n", "through the system twice, exposing them to different subsets of the\n", @@ -1907,7 +1907,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Given the realization that roughly 50% of the score seems to be\n", "‘subjective’ and 50% of the score seems to be ‘objective,’ then we can\n", @@ -2032,7 +2032,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -2175,7 +2175,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This notebook analyzes the reduction in reviewer confidence between\n", "reviewers that submit their reviews early and those that arrive late.\n", @@ -2830,7 +2830,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now we look at the actual impact of the papers published using the\n", "Semantic Scholar data base for tracking citations." @@ -3203,7 +3203,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Under the simple model we have outlined, we can be confident that there\n", "is inconsistency between two independent committees, but the level of\n", diff --git a/_notebooks/2021-07-07-ml-and-the-physical-world-trustworthy-ai.ipynb b/_notebooks/2021-07-07-ml-and-the-physical-world-trustworthy-ai.ipynb index ba13f718..60481f29 100644 --- a/_notebooks/2021-07-07-ml-and-the-physical-world-trustworthy-ai.ipynb +++ b/_notebooks/2021-07-07-ml-and-the-physical-world-trustworthy-ai.ipynb @@ -66,7 +66,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -77,7 +77,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Conway’s game of life is a cellular automata where the cells obey\n", "three very simple rules. The cells live on a rectangular grid, so that\n", @@ -243,7 +243,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Horton Conway, as the creator of the game of life, could be seen\n", "somehow as the god of this small universe. He created the rules. The\n", @@ -342,7 +342,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -358,7 +358,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -436,7 +436,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One project where a number of components of machine learning and the\n", "physical world come together is Amazon’s Prime Air drone delivery\n", @@ -615,7 +615,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "An example of a complex decision making system might be an automated\n", "buying system. In such a system, the idea is to match demand for\n", @@ -706,7 +706,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -727,7 +727,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To construct such complex systems an approach known as “separation of\n", "concerns” has been developed. The idea is that you architect your\n", @@ -757,7 +757,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -823,7 +823,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Supervised machine learning models are data-driven statistical\n", "functional estimators. Each ML model is trained to perform a task.\n", @@ -896,7 +896,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1013,7 +1013,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1069,7 +1069,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2021-07-13-ml-and-the-physical-world-tuebingen.ipynb b/_notebooks/2021-07-13-ml-and-the-physical-world-tuebingen.ipynb index 821d2f95..0d02346b 100644 --- a/_notebooks/2021-07-13-ml-and-the-physical-world-tuebingen.ipynb +++ b/_notebooks/2021-07-13-ml-and-the-physical-world-tuebingen.ipynb @@ -65,7 +65,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -76,7 +76,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Conway’s game of life is a cellular automata where the cells obey\n", "three very simple rules. The cells live on a rectangular grid, so that\n", @@ -242,7 +242,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Horton Conway, as the creator of the game of life, could be seen\n", "somehow as the god of this small universe. He created the rules. The\n", @@ -339,7 +339,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -383,7 +383,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -406,7 +406,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -422,7 +422,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -493,7 +493,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -535,7 +535,7 @@ "\n", "## Process Automation \n", "\n", - "\n", + "\n", "\n", "\n", "\n", @@ -566,7 +566,7 @@ "\n", "## Artificial Intelligence and Data Science \n", "\n", - "\n", + "\n", "\n", "Artificial intelligence has the objective of endowing computers with human-like intelligent capabilities. For example, understanding an image (computer vision) or the contents of some speech (speech recognition), the meaning of a sentence (natural language processing) or the translation of a sentence (machine translation).\n", "\n", @@ -612,7 +612,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One project where a number of components of machine learning and the\n", "physical world come together is Amazon’s Prime Air drone delivery\n", @@ -791,7 +791,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "An example of a complex decision making system might be an automated\n", "buying system. In such a system, the idea is to match demand for\n", @@ -882,7 +882,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -903,7 +903,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To construct such complex systems an approach known as “separation of\n", "concerns” has been developed. The idea is that you architect your\n", @@ -933,7 +933,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -999,7 +999,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Supervised machine learning models are data-driven statistical\n", "functional estimators. Each ML model is trained to perform a task.\n", @@ -1072,7 +1072,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1189,7 +1189,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1245,7 +1245,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2021-09-15-emulation.ipynb b/_notebooks/2021-09-15-emulation.ipynb index d0c9d527..8368b6af 100644 --- a/_notebooks/2021-09-15-emulation.ipynb +++ b/_notebooks/2021-09-15-emulation.ipynb @@ -65,7 +65,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "[John Horton Conway](https://en.wikipedia.org/wiki/John_Horton_Conway)\n", "was a mathematician who developed a game known as the Game of Life. He\n", @@ -86,7 +86,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Conway’s game of life is a cellular automata where the cells obey\n", "three very simple rules. The cells live on a rectangular grid, so that\n", @@ -261,7 +261,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Horton Conway, as the creator of the game of life, could be seen\n", "somehow as the god of this small universe. He created the rules. The\n", @@ -413,7 +413,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> The curve described by a simple molecule of air or vapor is regulated\n", "> in a manner just as certain as the planetary orbits; the only\n", @@ -474,7 +474,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "An example of a complex decision-making system might be a climate model,\n", "in such a system there are separate models for the atmosphere, the ocean\n", @@ -667,7 +667,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -742,7 +742,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Let $\\mathbf{ x}$ be a random variable defined over the real numbers,\n", "$\\Re$, and $f(\\cdot)$ be a function mapping between the real numbers\n", @@ -1130,7 +1130,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1307,7 +1307,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1529,7 +1529,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1603,7 +1603,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1926,7 +1926,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This introduction is based on [Introduction to Global Sensitivity\n", "Analysis with\n", @@ -2136,7 +2136,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The Ishigami function (Ishigami and Homma, 1989) is a well-known test\n", "function for uncertainty and sensitivity analysis methods because of its\n", @@ -2688,7 +2688,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2822,7 +2822,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now we will build an emulator for the catapult using the experimental\n", "design loop.\n", diff --git a/_notebooks/2021-11-04-deep-gaussian-processes-a-motivation-and-introduction.ipynb b/_notebooks/2021-11-04-deep-gaussian-processes-a-motivation-and-introduction.ipynb index 12e80678..34c6ee06 100644 --- a/_notebooks/2021-11-04-deep-gaussian-processes-a-motivation-and-introduction.ipynb +++ b/_notebooks/2021-11-04-deep-gaussian-processes-a-motivation-and-introduction.ipynb @@ -68,7 +68,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The fourth industrial revolution bears the particular hallmark of being\n", "the first revolution that has been named before it has happened. This is\n", @@ -108,7 +108,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -165,7 +165,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Any process of automation allows us to scale what we do by codifying a\n", "process in some way that makes it efficient and repeatable. Machine\n", @@ -251,7 +251,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classical statistical models and simple machine learning models have a\n", "great deal in common. The main difference between the fields is\n", @@ -290,7 +290,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -314,7 +314,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -387,7 +387,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -461,7 +461,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In January 2016, the UK company DeepMind’s machine learning system\n", "AlphaGo won a challenge match in which it beat the world champion Go\n", @@ -601,7 +601,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One view of Bayesian inference is to assume we are given a mechanism for\n", "generating samples, where we assume that mechanism is representing an\n", @@ -825,7 +825,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Even in the early days of Gaussian processes in machine learning, it was\n", "understood that we were throwing something fundamental away. This is\n", @@ -849,7 +849,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One potential problem is that as the number of nodes in two adjacent\n", "layers increases, the number of parameters in the affine transformation\n", @@ -910,7 +910,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -974,7 +974,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now if we replace each of these neural networks with a Gaussian process.\n", "This is equivalent to taking the limit as the width of each layer goes\n", @@ -998,7 +998,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "$$\\mathbf{ y}= \\mathbf{ f}_4\\left(\\mathbf{ f}_3\\left(\\mathbf{ f}_2\\left(\\mathbf{ f}_1\\left(\\mathbf{ x}\\right)\\right)\\right)\\right)$$\n", "\n", @@ -1086,7 +1086,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "If the result of composing many functions together is simply another\n", "function, then why do we bother? The key point is that we can change the\n", @@ -1203,7 +1203,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The [Planck space\n", "craft](https://en.wikipedia.org/wiki/Planck_(spacecraft)) was a European\n", @@ -1486,7 +1486,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Gaussian processes are a flexible tool for non-parametric analysis with\n", "uncertainty. The GPy software was started in Sheffield to provide a easy\n", @@ -1559,7 +1559,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1674,7 +1674,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1728,7 +1728,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Our first objective will be to perform a Gaussian process fit to the\n", "data, we’ll do this using the [GPy\n", @@ -1882,7 +1882,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Let’s see if a deep Gaussian process can help here. We will construct a\n", "deep Gaussian process with one hidden layer (i.e. one Gaussian process\n", @@ -2124,7 +2124,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Next we consider a simple step function data set." ] @@ -2251,7 +2251,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "First we initialize a deep Gaussian process with three latent layers\n", "(four layers total). Within each layer we create a GP with an\n", @@ -2508,7 +2508,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2563,7 +2563,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2579,7 +2579,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -2762,7 +2762,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We will look at a sub-sample of the MNIST digit data set.\n", "\n", @@ -2832,7 +2832,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3205,7 +3205,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3315,7 +3315,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2021-11-11-data-first-culture-post-digital-transformation-and-intellectual-debt.ipynb b/_notebooks/2021-11-11-data-first-culture-post-digital-transformation-and-intellectual-debt.ipynb index 7ac0fcbb..3f448748 100644 --- a/_notebooks/2021-11-11-data-first-culture-post-digital-transformation-and-intellectual-debt.ipynb +++ b/_notebooks/2021-11-11-data-first-culture-post-digital-transformation-and-intellectual-debt.ipynb @@ -94,7 +94,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -140,7 +140,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -186,7 +186,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -212,7 +212,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -223,7 +223,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -311,7 +311,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -368,7 +368,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -466,7 +466,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -565,7 +565,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -627,7 +627,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focussed on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -670,7 +670,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -747,7 +747,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> There are three types of lies: lies, damned lies and statistics\n", ">\n", @@ -906,7 +906,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -950,7 +950,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1041,7 +1041,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox is the modern phenomenon of “as we collect more\n", "data, we understand less.” It is emerging in several domains, political\n", @@ -1093,7 +1093,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox has a sister: the big model paradox. As we build\n", "more and more complex models, we start believing that we have a\n", @@ -1111,7 +1111,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -1189,7 +1189,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -1204,7 +1204,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -1287,7 +1287,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -1573,7 +1573,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", diff --git a/_notebooks/2022-04-26-post-digital-transformation-decision-making-and-intellectual-debt.ipynb b/_notebooks/2022-04-26-post-digital-transformation-decision-making-and-intellectual-debt.ipynb index fe98cf90..0392203c 100644 --- a/_notebooks/2022-04-26-post-digital-transformation-decision-making-and-intellectual-debt.ipynb +++ b/_notebooks/2022-04-26-post-digital-transformation-decision-making-and-intellectual-debt.ipynb @@ -122,7 +122,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -168,7 +168,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In Sheffield we created a suite of software tools for ‘Open Data\n", "Science.’ Open data science is an approach to sharing code, models and\n", @@ -218,7 +218,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -264,7 +264,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -290,7 +290,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -301,7 +301,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -389,7 +389,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -446,7 +446,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -591,7 +591,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -690,7 +690,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -776,7 +776,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> There are three types of lies: lies, damned lies and statistics\n", ">\n", @@ -935,7 +935,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -979,7 +979,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1079,7 +1079,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -1141,7 +1141,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focussed on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -1196,7 +1196,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1227,7 +1227,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "There’s been much recent talk about GDPR, much of it implying that the\n", "recent incarnation is radically different from previous incarnations.\n", @@ -1338,7 +1338,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Need to understand why you are processing personal data, for example see\n", "the ICO’s [Lawful Basis\n", @@ -1366,7 +1366,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1429,7 +1429,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Another distinction I find helpful when thinking about intelligence is\n", "the difference between reflexive actions and reflective actions. We are\n", @@ -1466,7 +1466,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox is the modern phenomenon of “as we collect more\n", "data, we understand less.” It is emerging in several domains, political\n", @@ -1518,7 +1518,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox has a sister: the big model paradox. As we build\n", "more and more complex models, we start believing that we have a\n", @@ -1536,7 +1536,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -1614,7 +1614,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -1629,7 +1629,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -1712,7 +1712,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -1998,7 +1998,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -2096,7 +2096,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -2107,7 +2107,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The DELVE Initiative was convened by the Royal Society early in the\n", "pandemic in response for a perceived need to increase provide policy\n", @@ -2140,7 +2140,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "- First contact *3rd April*\n", "- First meeting *7th April*\n", @@ -2242,7 +2242,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -2253,7 +2253,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Corporate culture turns out to be an important component of how you can\n", "react to digital transformation. Amazon is a company that likes to take\n", @@ -2346,7 +2346,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Any policy question can be framed in a number of different ways - what\n", "are the health outcomes; what is the impact on NHS capacity; how are\n", @@ -2364,7 +2364,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To improve communication, we need to ‘externalise cognition’: have\n", "objects that are outside our brains, are persistent in the real world,\n", @@ -2426,7 +2426,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "See the Gorilla *don’t* be the Gorilla.\n", "\n", diff --git a/_notebooks/2022-05-10-the-neurips-experiment-snsf.ipynb b/_notebooks/2022-05-10-the-neurips-experiment-snsf.ipynb index 434f2187..fab993c3 100644 --- a/_notebooks/2022-05-10-the-neurips-experiment-snsf.ipynb +++ b/_notebooks/2022-05-10-the-neurips-experiment-snsf.ipynb @@ -67,7 +67,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In 2014 the NeurIPS conference had 1474 active reviewers (up from 1133\n", "in 2013), 92 area chairs (up from 67 in 2013) and two program chairs,\n", @@ -99,7 +99,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Chairing a conference starts with recruitment of the program committee,\n", "which is usually done in a few stages. The primary task is to recruit\n", @@ -147,7 +147,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The instructions to reviewers for the 2014 conference are still\n", "available [online\n", @@ -317,7 +317,7 @@ "\n", "## Speculation \n", "\n", - "\n", + "\n", "\n", "With the help of [Nicolo Fusi](http://nicolofusi.com/), [Charles Twardy](http://blog.scicast.org/tag/charles-twardy/) and the entire Scicast team we launched [a Scicast question](https://scicast.org/#!/questions/1083/trades/create/power) a week before the results were revealed. The comment thread for that question already had [an amount of interesting comment](https://scicast.org/#!/questions/1083/comments/power) before the conference. Just for informational purposes before we began reviewing Corinna forecast this figure would be 25% and I forecast it would be 20%. The box plot summary of predictions from Scicast is below.\n", "\n", @@ -336,7 +336,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The results of the experiment were as follows. From 170 papers 4 had to\n", "be withdrawn or were rejected without completing the review process, for\n", @@ -461,7 +461,7 @@ "\n", "## Reaction After Experiment \n", "\n", - "\n", + "\n", "\n", "There seems to have been a lot of discussion of the result, both at the conference and on bulletin boards since. Such discussion is to be encouraged, and for ease of memory, it is worth pointing out that the approximate proportions of papers in each category can be nicely divided in to eighths as follows. Accept-Accept 1 in 8 papers, Accept-Reject 3 in 8 papers, Reject-Reject, 5 in 8 papers. This makes the statistics we've computed above: inconsistency 1 in 4 (25%) accept precision 1 in 2 (50%) reject precision 5 in 6 (83%) and agreed accept rate of 1 in 6 (20%). This compares with the accept rate of 1 in 4.\n", "\n", @@ -491,7 +491,7 @@ "\n", "## A Random Committee @ 25% \n", "\n", - "\n", + "\n", "\n", "The first context we can place around the numbers is what would have happened at the 'Random Conference' where we simply accept a quarter of papers at random. In this NIPS the expected numbers of accepts would then have been given as in Table \\ref{table-random-committee}.\n", "\n", @@ -818,7 +818,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Calibration of reviewers is the process where different interpretations\n", "of the reviewing scale are addressed. The tradition of calibration goes\n", @@ -829,7 +829,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In this note book we deal with reviewer calibration. Our assumption is\n", "that the score from the $j$th reviwer for the $i$th paper is given by $$\n", @@ -904,7 +904,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "``` python\n", "import cmtutils as cu\n", @@ -1045,7 +1045,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now we wish to predict the bias corrected scores for the papers. That\n", "involves considering a variable $s_{i,j} = f_i + e_{i,j}$ which is the\n", @@ -1066,7 +1066,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can now sample from this posterior distribution of bias-adjusted\n", "scores jointly, to get a set of scores for all papers. For this set of\n", @@ -1128,7 +1128,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Here is the histogram of the reviewer scores after calibration.\n", "\n", @@ -1247,7 +1247,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "For NeurIPS 2014 we experimented with duplicate papers: we pushed papers\n", "through the system twice, exposing them to different subsets of the\n", @@ -1357,7 +1357,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Given the realization that roughly 50% of the score seems to be\n", "‘subjective’ and 50% of the score seems to be ‘objective,’ then we can\n", @@ -1477,7 +1477,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One facet that we can explore is what the final fate of papers that are\n", "rejected by the conference is.\n", @@ -1642,7 +1642,7 @@ "\n", "## Effect of Late Reviews \n", "\n", - "\n", + "\n", "\n", "This notebook analyzes the reduction in reviewer confidence between\n", "reviewers that submit their reviews early and those that arrive late.\n", @@ -2302,7 +2302,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now we look at the actual impact of the papers published using the\n", "Semantic Scholar data base for tracking citations." @@ -2675,7 +2675,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Under the simple model we have outlined, we can be confident that there\n", "is inconsistency between two independent committees, but the level of\n", diff --git a/_notebooks/2022-06-06-deep-gaussian-processes-a-motivation-and-introduction-bristol.ipynb b/_notebooks/2022-06-06-deep-gaussian-processes-a-motivation-and-introduction-bristol.ipynb index b72e4851..287f9fcf 100644 --- a/_notebooks/2022-06-06-deep-gaussian-processes-a-motivation-and-introduction-bristol.ipynb +++ b/_notebooks/2022-06-06-deep-gaussian-processes-a-motivation-and-introduction-bristol.ipynb @@ -68,7 +68,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The fourth industrial revolution bears the particular hallmark of being\n", "the first revolution that has been named before it has happened. This is\n", @@ -108,7 +108,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -165,7 +165,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Any process of automation allows us to scale what we do by codifying a\n", "process in some way that makes it efficient and repeatable. Machine\n", @@ -251,7 +251,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classical statistical models and simple machine learning models have a\n", "great deal in common. The main difference between the fields is\n", @@ -290,7 +290,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -314,7 +314,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -387,7 +387,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -461,7 +461,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In January 2016, the UK company DeepMind’s machine learning system\n", "AlphaGo won a challenge match in which it beat the world champion Go\n", @@ -601,7 +601,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One view of Bayesian inference is to assume we are given a mechanism for\n", "generating samples, where we assume that mechanism is representing an\n", @@ -825,7 +825,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Even in the early days of Gaussian processes in machine learning, it was\n", "understood that we were throwing something fundamental away. This is\n", @@ -849,7 +849,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One potential problem is that as the number of nodes in two adjacent\n", "layers increases, the number of parameters in the affine transformation\n", @@ -910,7 +910,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -974,7 +974,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now if we replace each of these neural networks with a Gaussian process.\n", "This is equivalent to taking the limit as the width of each layer goes\n", @@ -998,7 +998,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "$$\\mathbf{ y}= \\mathbf{ f}_4\\left(\\mathbf{ f}_3\\left(\\mathbf{ f}_2\\left(\\mathbf{ f}_1\\left(\\mathbf{ x}\\right)\\right)\\right)\\right)$$\n", "\n", @@ -1086,7 +1086,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "If the result of composing many functions together is simply another\n", "function, then why do we bother? The key point is that we can change the\n", @@ -1203,7 +1203,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The [Planck space\n", "craft](https://en.wikipedia.org/wiki/Planck_(spacecraft)) was a European\n", @@ -1486,7 +1486,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Gaussian processes are a flexible tool for non-parametric analysis with\n", "uncertainty. The GPy software was started in Sheffield to provide a easy\n", @@ -1559,7 +1559,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1674,7 +1674,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1728,7 +1728,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Our first objective will be to perform a Gaussian process fit to the\n", "data, we’ll do this using the [GPy\n", @@ -1882,7 +1882,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Let’s see if a deep Gaussian process can help here. We will construct a\n", "deep Gaussian process with one hidden layer (i.e. one Gaussian process\n", @@ -2124,7 +2124,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Next we consider a simple step function data set." ] @@ -2251,7 +2251,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "First we initialize a deep Gaussian process with three latent layers\n", "(four layers total). Within each layer we create a GP with an\n", @@ -2508,7 +2508,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2563,7 +2563,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2579,7 +2579,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -2762,7 +2762,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We will look at a sub-sample of the MNIST digit data set.\n", "\n", @@ -2832,7 +2832,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3205,7 +3205,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3315,7 +3315,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2022-06-09-data-first-culture-post-digital-transformation-and-intellectual-debt-june-22.ipynb b/_notebooks/2022-06-09-data-first-culture-post-digital-transformation-and-intellectual-debt-june-22.ipynb index b53f2573..8e059ee0 100644 --- a/_notebooks/2022-06-09-data-first-culture-post-digital-transformation-and-intellectual-debt-june-22.ipynb +++ b/_notebooks/2022-06-09-data-first-culture-post-digital-transformation-and-intellectual-debt-june-22.ipynb @@ -65,7 +65,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -142,7 +142,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -192,7 +192,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In Sheffield we created a suite of software tools for ‘Open Data\n", "Science.’ Open data science is an approach to sharing code, models and\n", @@ -246,7 +246,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -296,7 +296,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -323,7 +323,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -335,7 +335,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -554,7 +554,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -612,7 +612,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -713,7 +713,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -813,7 +813,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -876,7 +876,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focussed on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -920,7 +920,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1011,7 +1011,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> There are three types of lies: lies, damned lies and statistics\n", ">\n", @@ -1173,7 +1173,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1220,7 +1220,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1325,7 +1325,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox is the modern phenomenon of “as we collect more\n", "data, we understand less.” It is emerging in several domains, political\n", @@ -1378,7 +1378,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox has a sister: the big model paradox. As we build\n", "more and more complex models, we start believing that we have a\n", @@ -1397,7 +1397,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -1480,7 +1480,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -1496,7 +1496,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -1587,7 +1587,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -1896,7 +1896,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1944,7 +1944,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "See the Gorilla *don’t* be the Gorilla.\n", "\n", diff --git a/_notebooks/2022-06-17-deep-gaussian-processes-a-motivation-and-introduction-sheffield.ipynb b/_notebooks/2022-06-17-deep-gaussian-processes-a-motivation-and-introduction-sheffield.ipynb index e8ec348c..773fa163 100644 --- a/_notebooks/2022-06-17-deep-gaussian-processes-a-motivation-and-introduction-sheffield.ipynb +++ b/_notebooks/2022-06-17-deep-gaussian-processes-a-motivation-and-introduction-sheffield.ipynb @@ -68,7 +68,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The fourth industrial revolution bears the particular hallmark of being\n", "the first revolution that has been named before it has happened. This is\n", @@ -108,7 +108,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -165,7 +165,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Any process of automation allows us to scale what we do by codifying a\n", "process in some way that makes it efficient and repeatable. Machine\n", @@ -251,7 +251,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classical statistical models and simple machine learning models have a\n", "great deal in common. The main difference between the fields is\n", @@ -290,7 +290,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -314,7 +314,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -387,7 +387,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -461,7 +461,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In January 2016, the UK company DeepMind’s machine learning system\n", "AlphaGo won a challenge match in which it beat the world champion Go\n", @@ -601,7 +601,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One view of Bayesian inference is to assume we are given a mechanism for\n", "generating samples, where we assume that mechanism is representing an\n", @@ -825,7 +825,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Even in the early days of Gaussian processes in machine learning, it was\n", "understood that we were throwing something fundamental away. This is\n", @@ -849,7 +849,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "One potential problem is that as the number of nodes in two adjacent\n", "layers increases, the number of parameters in the affine transformation\n", @@ -910,7 +910,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -974,7 +974,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now if we replace each of these neural networks with a Gaussian process.\n", "This is equivalent to taking the limit as the width of each layer goes\n", @@ -998,7 +998,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "$$\\mathbf{ y}= \\mathbf{ f}_4\\left(\\mathbf{ f}_3\\left(\\mathbf{ f}_2\\left(\\mathbf{ f}_1\\left(\\mathbf{ x}\\right)\\right)\\right)\\right)$$\n", "\n", @@ -1086,7 +1086,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "If the result of composing many functions together is simply another\n", "function, then why do we bother? The key point is that we can change the\n", @@ -1203,7 +1203,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The [Planck space\n", "craft](https://en.wikipedia.org/wiki/Planck_(spacecraft)) was a European\n", @@ -1486,7 +1486,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Gaussian processes are a flexible tool for non-parametric analysis with\n", "uncertainty. The GPy software was started in Sheffield to provide a easy\n", @@ -1559,7 +1559,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1674,7 +1674,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1728,7 +1728,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Our first objective will be to perform a Gaussian process fit to the\n", "data, we’ll do this using the [GPy\n", @@ -1882,7 +1882,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Let’s see if a deep Gaussian process can help here. We will construct a\n", "deep Gaussian process with one hidden layer (i.e. one Gaussian process\n", @@ -2124,7 +2124,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Next we consider a simple step function data set." ] @@ -2251,7 +2251,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "First we initialize a deep Gaussian process with three latent layers\n", "(four layers total). Within each layer we create a GP with an\n", @@ -2508,7 +2508,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2563,7 +2563,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -2579,7 +2579,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -2762,7 +2762,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We will look at a sub-sample of the MNIST digit data set.\n", "\n", @@ -2832,7 +2832,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3205,7 +3205,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3315,7 +3315,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", diff --git a/_notebooks/2022-07-07-data-first-culture-post-digital-transformation-and-intellectual-debt-july-22.ipynb b/_notebooks/2022-07-07-data-first-culture-post-digital-transformation-and-intellectual-debt-july-22.ipynb index 04bc56e4..4ffe2b2a 100644 --- a/_notebooks/2022-07-07-data-first-culture-post-digital-transformation-and-intellectual-debt-july-22.ipynb +++ b/_notebooks/2022-07-07-data-first-culture-post-digital-transformation-and-intellectual-debt-july-22.ipynb @@ -60,7 +60,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -132,7 +132,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -178,7 +178,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In Sheffield we created a suite of software tools for ‘Open Data\n", "Science.’ Open data science is an approach to sharing code, models and\n", @@ -228,7 +228,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -274,7 +274,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -300,7 +300,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -311,7 +311,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -399,7 +399,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -456,7 +456,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -554,7 +554,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -653,7 +653,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -715,7 +715,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focussed on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -758,7 +758,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -844,7 +844,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> There are three types of lies: lies, damned lies and statistics\n", ">\n", @@ -1003,7 +1003,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1047,7 +1047,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1147,7 +1147,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox is the modern phenomenon of “as we collect more\n", "data, we understand less.” It is emerging in several domains, political\n", @@ -1199,7 +1199,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox has a sister: the big model paradox. As we build\n", "more and more complex models, we start believing that we have a\n", @@ -1217,7 +1217,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -1295,7 +1295,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -1310,7 +1310,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -1393,7 +1393,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -1679,7 +1679,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1726,7 +1726,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "See the Gorilla *don’t* be the Gorilla.\n", "\n", diff --git a/_notebooks/2022-09-13-emulation-2022.ipynb b/_notebooks/2022-09-13-emulation-2022.ipynb index d7835513..f596849a 100644 --- a/_notebooks/2022-09-13-emulation-2022.ipynb +++ b/_notebooks/2022-09-13-emulation-2022.ipynb @@ -71,7 +71,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "[John Horton Conway](https://en.wikipedia.org/wiki/John_Horton_Conway)\n", "was a mathematician who developed a game known as the Game of Life. He\n", @@ -93,7 +93,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Conway’s game of life is a cellular automata where the cells obey\n", "three very simple rules. The cells live on a rectangular grid, so that\n", @@ -269,7 +269,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "John Horton Conway, as the creator of the game of life, could be seen\n", "somehow as the god of this small universe. He created the rules. The\n", @@ -424,7 +424,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> The curve described by a simple molecule of air or vapor is regulated\n", "> in a manner just as certain as the planetary orbits; the only\n", @@ -486,7 +486,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "An example of a complex decision-making system might be a climate model,\n", "in such a system there are separate models for the atmosphere, the ocean\n", @@ -682,7 +682,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -759,7 +759,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Let $\\mathbf{ x}$ be a random variable defined over the real numbers,\n", "$\\Re$, and $f(\\cdot)$ be a function mapping between the real numbers\n", @@ -1115,7 +1115,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We’re introducing you to the optimization and analysis of real-world\n", "models through emulation, this domain is part of a broader field known\n", @@ -1193,7 +1193,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1438,7 +1438,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ], "id": "d90a8747-af6b-4668-a11d-b4a409d5d78f" }, @@ -1674,7 +1674,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -1749,7 +1749,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ], "id": "0fe0f067-45bb-48b8-ab80-a70f471cffec" }, @@ -2103,7 +2103,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This introduction is based on [Introduction to Global Sensitivity\n", "Analysis with\n", @@ -2322,7 +2322,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The Ishigami function (Ishigami and Homma, 1989) is a well-known test\n", "function for uncertainty and sensitivity analysis methods because of its\n", @@ -2916,7 +2916,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3057,7 +3057,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Now we will build an emulator for the catapult using the experimental\n", "design loop.\n", @@ -3322,7 +3322,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "An example of a complex decision-making system might be an automated\n", "buying system. In such a system, the idea is to match demand for\n", @@ -3359,7 +3359,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The classical approach to building these systems was a ‘monolithic\n", "system’. Built in a similar way to the successful applications software\n", @@ -3517,7 +3517,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3544,7 +3544,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -3578,7 +3578,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "To construct such complex systems an approach known as “separation of\n", "concerns” has been developed. The idea is that you architect your\n", @@ -3609,7 +3609,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Zittrain points out the challenge around the lack of interpretability of\n", "individual ML models as the origin of intellectual debt. In machine\n", diff --git a/_notebooks/2022-11-10-data-first-culture-november-22.ipynb b/_notebooks/2022-11-10-data-first-culture-november-22.ipynb index 258afd8a..4904d415 100644 --- a/_notebooks/2022-11-10-data-first-culture-november-22.ipynb +++ b/_notebooks/2022-11-10-data-first-culture-november-22.ipynb @@ -60,7 +60,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -132,7 +132,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -178,7 +178,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In Sheffield we created a suite of software tools for ‘Open Data\n", "Science.’ Open data science is an approach to sharing code, models and\n", @@ -228,7 +228,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -274,7 +274,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -300,7 +300,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -311,7 +311,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -399,7 +399,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -456,7 +456,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Artificial intelligence has the objective of endowing computers with\n", "human-like intelligent capabilities. For example, understanding an image\n", @@ -554,7 +554,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -653,7 +653,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -715,7 +715,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focused on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -758,7 +758,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -844,7 +844,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "> There are three types of lies: lies, damned lies and statistics\n", ">\n", @@ -1002,7 +1002,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1045,7 +1045,7 @@ "\n", "\\[edit\\]" + "style=\"\">edit\\]" ] }, { @@ -1145,7 +1145,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox is the modern phenomenon of “as we collect more\n", "data, we understand less.” It is emerging in several domains, political\n", @@ -1197,7 +1197,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The big data paradox has a sister: the big model paradox. As we build\n", "more and more complex models, we start believing that we have a\n", @@ -1215,7 +1215,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -1293,7 +1293,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -1308,7 +1308,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -1391,7 +1391,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -1677,7 +1677,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "
\n", "\n", @@ -1724,7 +1724,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "See the Gorilla *don’t* be the Gorilla.\n", "\n", diff --git a/_notebooks/2022-11-14-how-engineers-solve-big-and-difficult-problems-part-1-the-challenge-opportunities-presented-to-engineers-by-ai-ml.ipynb b/_notebooks/2022-11-14-how-engineers-solve-big-and-difficult-problems-part-1-the-challenge-opportunities-presented-to-engineers-by-ai-ml.ipynb index 7a5aab46..7c3838b2 100644 --- a/_notebooks/2022-11-14-how-engineers-solve-big-and-difficult-problems-part-1-the-challenge-opportunities-presented-to-engineers-by-ai-ml.ipynb +++ b/_notebooks/2022-11-14-how-engineers-solve-big-and-difficult-problems-part-1-the-challenge-opportunities-presented-to-engineers-by-ai-ml.ipynb @@ -95,7 +95,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "This small package is a helper package for various notebook utilities\n", "used\n", @@ -145,7 +145,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "In Sheffield we created a suite of software tools for ‘Open Data\n", "Science’. Open data science is an approach to sharing code, models and\n", @@ -199,7 +199,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The `mlai` software is a suite of helper functions for teaching and\n", "demonstrating machine learning algorithms. It was first used in the\n", @@ -249,7 +249,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "As an exercise in understanding complexity, watch the following video.\n", "You will see the basketball being bounced around, and the players\n", @@ -332,7 +332,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We are going to see how inattention biases can play out in data analysis\n", "by going through a simple example. The analysis involves body mass index\n", @@ -348,7 +348,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The BMI Steps example is taken from Yanai and Lercher (2020). We are\n", "given a data set of body-mass index measurements against step counts.\n", @@ -439,7 +439,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "We can also separate out the means from the male and female populations.\n", "In python this can be done by setting male and female indices as\n", @@ -684,7 +684,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "What is machine learning? At its most basic level machine learning is a\n", "combination of\n", @@ -742,7 +742,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Machine learning technologies have been the driver of two related, but\n", "distinct disciplines. The first is *data science*. Data science is an\n", @@ -796,7 +796,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "The high bandwidth of computers has resulted in a close relationship\n", "between the computer and data. Large amounts of information can flow\n", @@ -859,7 +859,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "Classically the field of statistics focused on mediating the\n", "relationship between the machine and the human. Our limited bandwidth of\n", @@ -903,7 +903,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n", @@ -961,7 +961,7 @@ "\n", "\\[edit\\]\n", + "style=\"\">edit\\]\n", "\n", "\n", "\n",