Towards EEG signals codification using contrastiveloss

Abstract

In this work we explore the use of contrastive learning to obtain vectorial representation of EEG signals. As contrastive learning is a self-supervised method, we propose the use of consecutive segments up to a certain window of time to define a property that will guide the contrastive learning. The results are promising given the small sample of EEG signals that we have access.

Tasks

Reproductions