A slight modification of a vanilla autoencoder, it aims to reconstruct an image with some noise added (one or two image quadrants removed randomly). Here, trained and evaluated on fashionMNIST dataset.
michalinabijak/autoencoder
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|