Multi-View Bayesian Correlated Component Analysis

TL;DR

This paper introduces Bayesian correlated component analysis, a hierarchical probabilistic model that assesses the degree of similarity in multi-view brain data, bridging canonical correlation and correlated component analysis.

Contribution

The paper presents a novel Bayesian model that infers the level of universality in multi-view data, extending existing methods to handle varying degrees of shared representations.

Findings

Favorable performance against three algorithms in simulated data.

Validated model on EEG dataset to assess variability across subjects.

Effectively bridges CCA and correlated component analysis.

Abstract

Correlated component analysis as proposed by Dmochowski et al. (2012) is a tool for investigating brain process similarity in the responses to multiple views of a given stimulus. Correlated components are identified under the assumption that the involved spatial networks are identical. Here we propose a hierarchical probabilistic model that can infer the level of universality in such multi-view data, from completely unrelated representations, corresponding to canonical correlation analysis, to identical representations as in correlated component analysis. This new model, which we denote Bayesian correlated component analysis, evaluates favourably against three relevant algorithms in simulated data. A well-established benchmark EEG dataset is used to further validate the new model and infer the variability of spatial representations across multiple subjects.