面向生物信息感知网络稀疏脑电测量的模糊粗糙情绪识别

Fuzzy rough emotion recognition based on sparse EEG sensing in biosensor network

情绪健康与人们的工作生活乃至社会公共安全紧密相关。情绪识别通过测量表征情绪状态的生物信息识别人的个体情绪，为情绪健康状态辨识提供依据。生物信息感知网络可用于复杂环境生物信息的感知测量，对特定场景下的情绪监测具有重要意义。本文提出一种面向生物信息感知网络稀疏脑电测量的模糊粗糙情绪识别方法，采用稀疏脑电测量设备以及无线可穿戴生物传感节点构建多模生物信息感知网络，测量提取情绪相关信息，并对多模生物信息进行融合分析，针对情绪本身的模糊粗糙特性、依据脑电专注度模糊门限提出一种改进的模糊粗糙近邻分类算法（FRNN）。该方法削减了28．20％的运算量，提高了情绪识别效率；同时减少了无关情绪样本对分类准确率的影响，提高情绪识别准确率6％～7％，识别率65．53％高于同类研究成果。本文在可穿戴网络架构下实现了情绪的快速识别，可为日常情绪健康监测提供有效参考依据。

Emotion recognition is a process in which emotions are identified and recognized according to related bio-signals.This paper develops a fuzzy rough emotion recognition method based on sparse EEG sensing in biosensor network,which can help monitor emotional health in specific environment.Sparse EEG sensing equipment and wireless wearable bio-sensing nodes are used to record EEG data and multimodal bio-signals under different stimulations.According to the fuzzy and rough characteristics of emotion recognition,this paper proposes an adapted Fuzzy Rough Nearest Neighbors （FRNN）classification method based on the data fusion of sparse EEG and bio-sig-nals.A fuzzy threshold of EEG attention value is proposed to reduce 28.20%of the calculation task and improve the accuracy by 6%~7%,which reaches to 65.53%（higher than related works）.The experiment results indicate that this framework realizes fast and relia-ble emotion recognition in wearable web-enabled sensing environment and provides a solution to primary diagnosis of emotional health.