基于动态自适应策略的SSVEP快速目标选择方法

High-speed target selection method for SSVEP based on a dynamic stopping strategy

基于稳态视觉诱发电位（SSVEP）的脑-机接口（BCI）系统具有高速、大指令集等优点,其信息传输率（ITR）的进一步提升对系统走向实际应用具有重要意义。该文采用包含35名被试的公开数据集,使用任务相关成分分析的分类模型识别脑电图数据中的SSVEP成分,进而运用基于Bayes估计的动态自适应策略评估分类结果的置信度。实验结果表明：动态自适应策略所得到的平均ITR（230b/min）比传统的静态自适应策略（204b/min）提升了12.7%,基于Bayes的动态自适应策略可以进一步提升SSVEP-BCI的性能。

The steady-state visual evoked potential（SSVEP）based brain-computer interfaces（BCIs）are very fast and have huge instruction sets.However,the information transfer rate（ITR）needs to be increased for practical applications.A classification model of the task-relevant component analysis is used on a public data set of 35 subjects to recognize the SSVEP component in the electroencephalography data.A dynamic stopping strategy based on Bayes estimation is used to evaluate the confidence of the classification results.The results show that the dynamic stopping strategy（230 b/min）improves the average ITR by 12.7%compared with the conventional fixed stopping strategy（204 b/min）.Thus,this result shows how SSVEP-BCI can be further improved by the Bayes-based dynamic stopping strategy.