My best attempt to construct a lovely code that implement Autoencoders Neural Networks.

If you want to use this code, after downloading this repo, make sure to have all the required libraries, that are:

I suggest you to configure an ad hoc conda environment.

After loading your datasets in the directory dataset and after specifing your training parameters of interest in the file src/params.py, you can simply start the training using the command python autoencoder.py. The program will print on a file in the results dir the average and the standard deviation of the losses computed over train set, validation set (according to the external cross validation procedure) and test set (if test set is present and you want to print also this observables).