Convolutional Autoencoder and Deep Infomax for epileptic EEG signals clustering. This work of thesis wants to show how Convolutional Autoencoder and Deep Infomax networks can help in the diagnosis of epilepsy, because nowadays the duty belongs to five doctors. Different tests have been effectuated using two different networks and applying different initial conditions such as: epoch length, number of channel observed and different composition of traning set. Unluckily, hoped results have not been achieved, but important starting points for furhter developments are presented.