AutoEncoder

class deepinv.models.AutoEncoder(dim_input, dim_mid=1000, dim_hid=32, residual=True)

Bases: Denoiser

Simple fully connected autoencoder network.

Simple architecture that can be used for debugging or fast prototyping.

Parameters:

• dim_input (int) – total number of elements (pixels) of the input.

• dim_hid (int) – number of features in intermediate layer.

• dim_hid – latent space dimension.

• residual (int) – use a residual connection between input and output.

forward(x, sigma=None, **kwargs)

Applies denoiser \(\denoiser{x}{\sigma}\).

Parameters:

• x (torch.Tensor) – noisy input.

• sigma (torch.Tensor, float) – noise level.

Returns:

(torch.Tensor) Denoised tensor.