Cross-Subject Transfer Learning Improves the Practicality of Real-World Applications of Brain-Computer Interfaces
share
Steady-state visual evoked potential (SSVEP)-based brain computer-interfaces (BCIs) have shown its robustness in achieving high information transfer rate. State-of-the-art training-based SSVEP decoding methods such as extended Canonical Correlation Analysis (CCA) and Task-Related Component Analysis (TRCA) are the major players that elevate the efficiency of the SSVEP-based BCIs through a calibration process. However, because of large human variability both across individuals and within individuals over time, we need to collect calibration (training) data from each individual before each session, which could be laborious and time-consuming, deteriorating its practicality in a real-world context. This study aims to develop a cross-subject transferring approach to reduce the need for training data from a test user. Study results showed that a new least-squares transformation (LST) method was able to significantly reduce the training templates required for a 40-class SSVEP BCI. The LST method may lead to numerous real-world applications of a plug-and-play high-speed SSVEP-based BCI.
READ FULL TEXT
