Learning representations from EEG with deep recurrent-convolutional neural networks

[jinglescode]

Paper

Link: https://arxiv.org/pdf/1511.06448.pdf
Year: 2016

Summary

• designed to preserve the spatial, spectral, and temporal structure of EEG which leads to finding features that are less sensitive to variations and distortions within each dimension
• robust to inter- and intra-subject differences, as well as to measurementrelated noise

Contributions and Distinctions from Previous Works

• none of these studies attempted to jointly preserve the structure of EEG data within space, time, and frequency

Methods

• transform EEG activities into a sequence of topology-preserving multi-spectral images
• train a deep recurrent convolutional network inspired by state-of-the-art video classification techniques to learn robust representations from the sequence of images
• working memory experiment. 15 subjects, 64 electrodes, sampling 500 Hz, 240 trial per subject, 3.5 seconds per trial, 4 categories

Results

conv + LSTM + 1D conv performed best

Code

https://github.com/VDelv/EEGLearn-Pytorch