A Comparison of ERPs, SVM Decoding, and Cross Mahalanobis Distance Analysis in EEG
Date:
Presented to the UC Davis EEG Consortium, a group of researchers meeting monthly to discuss topics in human electrophysiology. This talk shared preliminary results comparing traditional ERP averaging methods to multivariate pattern analysis (MVPA) techniques, including support vector machine (SVM) decoding and cross-validated Mahalanobis distance.
Emphasized the potential advantages of applying MVPA to ERP datasets in order to detect subtle condition differences with increased sensitivity. The discussion focused on the implications for future EEG analysis pipelines.