An Adaptive Task-Related Component Analysis Method for SSVEP recognition

TL;DR

This paper introduces an adaptive, data-driven spatial filtering method for SSVEP recognition that effectively utilizes limited calibration data through a multitask Bayesian learning approach, significantly improving detection performance.

Contribution

It presents a novel adaptive spatial filtering technique leveraging multitask Bayesian learning to enhance SSVEP detection with scarce calibration data.

Findings

Outperforms existing methods on benchmark datasets

Significantly improves SSVEP detection accuracy

Effective with limited calibration trials

Abstract

Steady-state visual evoked potential (SSVEP) recognition methods are equipped with learning from the subject's calibration data, and they can achieve extra high performance in the SSVEP-based brain-computer interfaces (BCIs), however their performance deteriorate drastically if the calibration trials are insufficient. This study develops a new method to learn from limited calibration data and it proposes and evaluates a novel adaptive data-driven spatial filtering approach for enhancing SSVEPs detection. The spatial filter learned from each stimulus utilizes temporal information from the corresponding EEG trials. To introduce the temporal information into the overall procedure, an multitask learning approach, based on the bayesian framework, is adopted. The performance of the proposed method was evaluated into two publicly available benchmark datasets, and the results demonstrated that our method outperform competing methods by a significant margin.