Learning across multi-stimulus enhances target recognition methods in SSVEP-based BCIs

摘要

Objective. Latest target recognition methods that are equipped with learning from the subject’scalibration data, represented by the extended canonical correlation analysis (eCCA) and theensemble task-related component analysis (eTRCA), can achieve extra high performance in thesteady-state visual evoked potential (SSVEP)-based brain–computer interfaces (BCIs), however theirperformance deteriorate drastically if the calibration trials are insufficient. This paper develops a newscheme to learn from limited calibration data. Approach. A learning across multiple stimuli schemeis proposed for the target recognition methods, which applies to learning the data correspondingto not only the target stimulus but also the other stimuli. The resulting optimization problems canbe simplified and solved utilizing the prior knowledge and properties of SSVEPs across differentstimuli. With the new learning scheme, the eCCA and the eTRCA can be extended to the multistimuluseCCA (ms-eCCA) and the multi-stimulus eTRCA (ms-eTRCA), respectively, as well asa combination of them (i.e. ms-eCCA+ms-eTRCA) that incorporates their merits. Main results.Evaluation and comparison using an SSVEP-BCI benchmark dataset with 35 subjects show that thems-eCCA (or ms-eTRCA) performs significantly better than the eCCA (or eTRCA) method whilethe ms-eCCA+ms-eTRCA performs the best. With the learning across stimuli scheme, the existingtarget recognition methods can be further improved in terms of the target recognition performanceand the ability against insufficient calibration. Significance. A new learning scheme is proposedtowards the efficient use of the calibration data, providing enhanced performance and savingcalibration time in the SSVEP-based BCIs.