Keras: How to choose the loss function for Binary Autoencoders

[kingfengji]

Hi all,

I wish to train a stacked auto-encoder, but the input vectors* are binary, ie [0,0,1,1,0,1..1], and they are quite sparse......*

so picking the loss function as mse doesnt work so well, I wish to know
1) will it makes sense changing the last layer of decoder into a softmax activation *
2) which *loss function should I use.

Thanks

[lemuriandezapada]

binary_crossentropy

[kingfengji]

@lemuriandezapada
no, it is not a one-hot vector, it has multiple ones in it.

[pengpaiSH]

@kingfengji How did you finally fix that problem?

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed after 30 days if no further activity occurs, but feel free to re-open a closed issue if needed.