I'm using Autoencoder for feature extracting. I stuck with how to choose good number of dimension of encoder layer (latent layer). After training dataset, the model gave the latent layer (embedding layer) with some zero value in the vector result.

For example, the embedding layer have 4 dimensions, one of node (unit) in embedding layer has value [0.67 0.0 2.13 0.43]. That I suppose they should 4 values different zero value.

I think my problem that I choose too many dimension for embedding layer which is not necessary and should smaller dimension, such as 3 or 2 dimensions.

So, my question how to choose good size for embedding layer?


Treat it as a hyperparameter, and do a hyperparameter search over it. This requires having a good metric for your model performance. Which usually means having a validation set with labels. Training can still be unsupervised.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.