Fig. 6
From: Detecting epileptic seizures with electroencephalogram via a context-learning model
An example of EEG sequence translation. Each EEG fragment can be decomposed into a combination of EEG words in the EEG dictionary. Each EEG word corresponds to a hidden unit of the sparse autoencoder in the dictionary learning step, and the distribution of EEG words is the normalized weights of the hidden units. And then an EEG fragment is sampled to an EEG word according the word distribution