Riemannian classification of EEG signals with missing values

TL;DR

This paper introduces a Riemannian classification method for EEG signals with missing data, utilizing an expectation-maximization approach to improve accuracy over existing methods in brain-computer interface tasks.

Contribution

It presents a novel strategy using observed-data likelihood and EM algorithm for classifying EEG signals with missing values, outperforming state-of-the-art techniques.

Findings

Proposed method outperforms existing approaches on real EEG data.

The approach is applicable to various missing data scenarios.

It improves classification accuracy in brain-computer interface tasks.

Abstract

This paper proposes a strategy to handle missing data for the classification of electroencephalograms using covariance matrices. It relies on the observed-data likelihood within an expectation-maximization algorithm. This approach is compared to two existing state-of-the-art methods: (i) covariance matrices computed with imputed data; (ii) Riemannian averages of partially observed covariance matrix. All approaches are combined with the minimum distance to Riemannian mean classifier and applied to a classification task of two widely known paradigms of brain-computer interfaces. In addition to be applicable for a wider range of missing data scenarios, the proposed strategy generally performs better than other methods on the considered real EEG data.