猜拳博弈过程脑电的时频特征分析及基于优选特征的博弈决策预测

发布时间：2018-02-03 05:02

本文关键词： 优选特征脑活动博弈过程猜拳脑电波　出处：《信号处理》2017年03期 　论文类型：期刊论文

【摘要】：博弈是人类根据信息和经验做出使利益最大化决策的一种行为。对博弈的脑研究由来已久,使用的方法也各不相同,但都侧重于博弈的原理研究,对博弈过程中脑活动的认知规律研究几乎无人涉足。本文设计了具有静息、评估、决策、反馈、休息五个阶段的"石头-剪刀-布"猜拳博弈实验范式。采集并提取了17名被试在博弈活动不同阶段脑电波的时频特征,通过研究评估阶段脑活动的认知规律,提出了基于优选特征的博弈决策预测方法。使用SVM对决策过程训练并识别,达到了80%的识别率,说明了本文提出的优选特征能够很好地表达博弈认知过程的特点。
[Abstract]:Game is a kind of behavior that human make the most benefit decision according to information and experience. The brain research on game has been for a long time, the methods used are different, but they all focus on the principle of the game. Almost no one has set foot in the study of the cognitive law of brain activity in the process of game. This paper designs a design with rest, evaluation, decision-making, feedback. The experiment paradigm of "stone, scissors and cloth" game of guessing fist in five stages of rest was used to collect and extract the time-frequency characteristics of brain waves of 17 subjects in different stages of game activities. By studying the cognitive law of brain activity in evaluation stage, a game decision prediction method based on optimal selection features is proposed. The decision process is trained and recognized by SVM, and the recognition rate of 80% is achieved. The results show that the optimal selection features can well express the characteristics of game cognitive process.
【作者单位】：哈尔滨工业大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(61171186,61271345,61671187) 深圳市基础研究项目(JCYJ20150929143955341) 语言语音教育部-微软重点实验室开放基金资助项目(HIT.KLOF.20150XX,HIT.KLOF.20160xx) 中央高校基本科研业务费专项资金(HIT.NSRIF.2012047)
【分类号】：R338;O225
【正文快照】： 1引言博弈是人类根据自己所掌握的信息和以往的生活经验,尽可能的做出使自己利益最大化决策的一种行为。博弈过程是在一定条件下,遵守一定的规则,依靠所掌握的信息,各自选择行为或策略并加以实施,取得收益的过程。博弈的精髓在于,在遵守规则的前提下最大化自己的利益。博弈过

【相似文献】