Reviews: Targeting EEG/LFP Synchrony with Neural Nets

The manuscript introduces the novel CNN named SyncNet, which was designed to detect synchrony/spectral coherence in brain electrophysiological signals. To overcome common issues in EEG (varying setup configuration, correlated channels, ...) a Gaussian Process Adaptor is used to transform EEG data into a pseudo input space. The design of SyncNet is chosen to allow interpretation of learned featruers. The method was applied to several publicly available datasets and its performance compared to state of the art algorithms. Results suggest that SyncNet yields competitive results.