摘要
博弈认知状态由于其严密的逻辑性和数学框架,是脑机接口的新兴领域。近年来受到数学家、心理学家和计算机学者广泛的关注,并且已经取得一定的研究成果。首先,本文设计了具有典型博弈活动的"石头-剪刀-布"认知状态研究实验。接着对博弈活动进行了阶段性的划分。然后,采集了17个被试的脑电数据,接着对采集到的脑电数据进行:(1)脑电预览;(2)伪迹剔除与矫正;(3)数字滤波;(4)转参考;(5)Double-ICA去伪迹;(6)脑电分段;(7)基线矫正等离线预处理操作。在此基础上,分别从统计学特征、时域特征和频域特征3个方面对被试出石头、剪刀、布3种情况下的脑电信号进行特征提取,使用基于关联的特征选择方法进行特征选择。最后使用SVM分类器进行分类识别,其中SVM的核函数采用线性核函数,达到了83.3%的识别率。
Game cognitive state,due to its rigorous logic and mathematical framework,is an emerging field of brain-computer interface. In recent years,it has drawn much attention from mathematicians,psychologists and computer scholars,and has achieved some research results. At first,this paper designs a cognitive state research experiment "stone-scissors-cloth",which is a typical game. Secondly,the paper divides the game into different stages. Then,the paper collects 17 subjects of EEG data,and with these EEG data,respectively performs the following steps:( 1) electrical preview;( 2) artifact correction;( 3) filter;( 4) reference;( 5)Double-ICA artifact correction;( 6) epoch;( 7) baseline correct offline preprocessing operations. On this basis,the paper extracts the features of EEG from stone,scissor and cloth in terms of statistics,time domain and frequency domain; next,selects features by using feature selection method based on correlation. Finally,the research uses the SVMclassifier for classification recognition,with linear kernel function as the SVMkernel function,and the recognition rate reaches 83.3%.
作者
刘新磊
李海峰
马琳
LIU Xinlei;LI Haifeng;MA Lin(School of Computer Science and Technology,Harbin Institute of Technology,Harbin 150001,China)
出处
《智能计算机与应用》
2018年第5期75-78,82,共5页
Intelligent Computer and Applications
基金
国家自然科学基金(61171186)
关键词
博弈
统计特征
功率谱
留一法
game
the statistical characteristics
power spectrum
leave-one-out method