An Asynchronous Detection Algorithm for SSVEP-Based BCI Using Gradient Boosting Decision Tree

Abstract

Asynchronous Brain-computer Interface (BCI) plays an essential role in practical applications, for it can detect intentional control (IC) states and non-control (NC) states directly, allowing users to send commands when they intend to do so. In this study, to achieve an efficient asynchronous BCI system, Gradient Boosting Decision Tree (GBDT) is applied to detect IC and NC states for the first time. Specifically, the steady-state visual evoked potentials (SSVEP) is chosen as the BCI paradigm. With the help of appropriate feature selection and optimization, the proposed method not only improved the recognition accuracy but also reduced the computational cost.