This project is the implementation of the following assigment. Click here to know more and Here to read the report.
We want to implement two of the Machine Learning models saw in class.
The first one is the Naive Bayes model, a supervised machine learningalgorithm. This algorithm use Bayes Theorem to generate a model.
We think that, since the images are centered and relatively simple we can say that, for example, if the center is not black the probability of the digitto be 0 (zero) should be low. Following the same logic, if the center is white the probability of the digit to be 1 (one) should be higher than the probability of the digit to be 0 (zero). Thinking in that way Naive Bayes model could work well also considering the fact that it assumes that all input variables are independent. If that assumption is not correct, then it could impact the accuracy of the model.
Thinking in that way we think Naive Bayes model could work well, known that is not the optimal choice.
The second one is a Neural Network model.
We choose this model becuase of the flexibility with different kind of problem and also their reliability with tasks involving a very large number of features.