charlesnord/convolutional-autoencoder

convolutional-autoencoder

cae

Re-implementation of code in this link: https://github.com/XifengGuo/DCEC/blob/master/ConvAE.py

With little modification, sigmoid activation function is added on the top of model, thus we can output images with pixel vaule between 0 and 1, and we can apply the binary crossentropy loss:

Here the crossentropy is averaged by the number of pixels, just like mse loss.

cae2

Use fully convolutional layer instead of fully connected layer.(i.e. removed the steps of reshaping feature maps).

The bottle neck features are in 2D space, so that we can directly visualize them, but the performance is not satisfying, after 20 epochs of iteration, we got the following clustering result and reconstruction result.

2D-CAE reconstruction

2D-CAE visualization

cae tsne

Encode the image to 10D vector, and use tsne to visualize it

10D-CAE reconstruction

10D-CAE tsne visualization

cvae

Convolutional variational autoencoder Encode the image to 10D vector, and visualize it by tsne

10D-CVAE reconstruction

10D-CVAE tsne visualization

conditional_conv_vae

Modified based on convolutional variational autoencoder, I tested only input label in decoder and input label both in encoder and decoder, here is the result.

input label only in decoder

input label in both encoder and decoder

We could easily find that the second is better, and this is what most conditional vae algorithms do.