I have a classification problem for which a feedforward, fully connected neural net works reasonably well (two classes, true positive and true negative rate close to 80%).

I want to get these rates to 90%, and more features is one of the catalysts for improvements I can think of.

Do autoencoders to learn additional, interesting features work well for problems that do not involve images?

  • $\begingroup$ Which programming language do you use? Did you have the code? If possible could you send to me, please? $\endgroup$
    – Mohammad
    Aug 30, 2016 at 15:03

1 Answer 1


Yes, but no-one can tell if they will work well for your problem, so just try it and see. Don't give up if it does not work at first, because training neural networks requires some practice; there are lots of parameters, and not every configuration will work well. Even the optimization algorithm is a hyperparameter.

  • $\begingroup$ thanks a lot Emre, will go ahead and try. Any particular library you recommend? thanks again. $\endgroup$ Jan 31, 2016 at 23:02
  • $\begingroup$ keras.io/layers/core/#autoencoder $\endgroup$
    – Emre
    Feb 1, 2016 at 4:05
  • $\begingroup$ awesome! I have already used keras, will try out the autoencoder $\endgroup$ Feb 1, 2016 at 10:36

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.