1D Transposed Convolutional Layer

[lars-reimann]

Is your feature request related to a problem?

We only have 2D transposed convolutions so far. 1D transposed convolutions can also be useful, e.g. for time series.

Desired solution

• Add a new class safeds.ml.nn.layers.ConvolutionalTranspose1DLayer.
• Superclass: Convolutional1DLayer (see 1D Convolutional Layer #853)
• Constructor parameters:
  • output_channel: int
  • kernel_size: int
  • keyword-only stride: int = 1
  • keyword-only padding: int = 0
  • keyword-only output_padding: int = 0
  • Call _check_bounds to ensure that:
    • output_channel >= 1
    • kernel_size >= 1
    • stride >= 1
    • padding >= 0
    • output_padding >= 0
• Implementation:
  • Use PyTorch's ConvTranspose1D internally

Possible alternatives (optional)

No response

Screenshots (optional)

No response

Additional Context (optional)

No response