基于脑电信号分析的情绪识别研究进展

Advances in Emotion Recognition Based on EEG Signal Analysis

情绪识别是一种通过分析生理信号来识别和理解人类情绪状态的技术,被广泛应用于临床辅助诊断治疗、社交媒体及人机交互等方面。相比于其他非生理信号,利用脑电图(EEG)进行情绪识别可以获取更为客观直接的情绪数据。本文分析了近年来国内外文献的相关研究,从情绪数据来源与预处理、特征提取、基于深度学习算法的识别模型和临床应用4个方面归纳梳理了情绪识别的研究进展。其中,特征提取与基于深度学习的算法识别在脑电信号分析和情绪识别中发挥了重要作用。最后,本文总结了基于脑电图情绪识别的临床应用研究现状;并从解决数据丢失问题和建立有效算法等方面对今后情绪识别的研究方向提出了一些展望。

Emotion recognition is a technique for recognizing and understanding human emotional states by analyzing physiological signals,and is widely used in clinically assisted diagnosis and treatment,social media,and human-computer interaction.Compared with other non-physiological signals,emotion recognition using electroencephalogram(EEG)can obtain more objective and direct emotion data.In this paper,we analyze the related research in domestic and international literature in recent years,and summarize the research progress of emotion recognition from four aspects,emotion data source and preprocessing,feature extraction,deep learning algorithm-based recognition model and clinical application.Among them,feature extraction and recognition based on deep learning algorithm play an important role in EEG signal analysis and emotion recognition.Finally,paper summarizes the clinical application and research status of emotion recognition based on EEG;and puts forward the outlook of future emotion recognition research in terms of solving the problem of data loss and establishing effective algorithms.