%0 Journal Article
%T 基于动态自适应策略的SSVEP快速目标选择方法<br>High-speed target selection method for SSVEP based on a dynamic stopping strategy
%A 王春慧
%A 江京
%A 李海洋
%A 许敏鹏
%A 印二威
%A 明东
%J 清华大学学报（自然科学版）
%D 2018
%R 10.16511/j.cnki.qhdxxb.2018.22.038
%X 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（230 b/min）比传统的静态自适应策略（204 b/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。<br>Abstract：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.
%K 脑-机接口
%K 稳态视觉诱发电位
%K 脑电图
%K 动态自适应策略
%K 任务相关成分分析
%K &lt
%K br&gt
%K brain-computer interface
%K steady-state visual evoked potential
%K electroencephalography
%K dynamic stopping strategy
%K task-related component analysis
%U http://jst.tsinghuajournals.com/CN/Y2018/V58/I9/788