基于在线Infomax算法的视觉诱发电位提取

EXTRACTION OF VISUAL EVOKED POTENTIALS BASED ON ON-LINE INFOMAX ALGORITHM

在线信息极大 (Infomax)算法是基于单次观测样本的盲源分离算法 ,具有实时自适应跟踪混合系统的时变特性的能力。针对算法易出现稳态失调的问题 ,结合二阶和四阶统计去相关的混合学习规则 ,采用自适应步长学习技术 ,提出了一种改进的在线Infomax算法 ,并将其用于视觉诱发电位 (VEP)的实时提取。实验结果表明 ,该方法改善了原有算法的收敛性能 ,盲源分离效果良好 ,并可在减少视觉刺激次数的前提下 ,有效地实现VEP信号增强和参数提取。

On-line Information maximization(Infomax) algorithm is a kind of blind signal separation based on single observation sample, it can track the real-time change of mixing system adaptively. To overcome the problem of misadjustment in the steady state, an improved on-line Infomax algorithm with hybrid learning rule of 2nd order and 4th order statistical decorrelations, and adaptive step size technique is presented in this paper. This algorithm is applied to the real-time extraction of Visual Evoked Potentials(VEP). Simulation and application experiment results show that the proposed algorithm can improve the convergence performance and effect of blind signal separation. The reinforcing and parameter extraction of VEP can be realized effectively on the premise of fewer vision stimulations.