Abstract: Dimensionality reduction and the automatic learning of key features from electroencephalographic (EEG) signals have always been challenging tasks. Variational autoencoders (VAEs) have been ...