In the recipe we will use the Greedy layer-wise approach to train the stacked autoencoders; to make the task easier, we will use shared weights, thus the corresponding encoder/decoder weights would be the transpose of each other.
In the recipe we will use the Greedy layer-wise approach to train the stacked autoencoders; to make the task easier, we will use shared weights, thus the corresponding encoder/decoder weights would be the transpose of each other.