基于混合智能优化算法的生理信号情感识别

doi:10.3969/j.issn.1000-0801.2010.09.034

摘要/Abstract

摘要：

让计算机具有识别情感的能力是情感智能的主要标志和实现高级别人机交互的重要前提，其中通过记录和分析生理信号来识别情感状态已经成为情感计算和人机交互研究领域中的热点。针对多生理信号情感识别过程中的特征冗余以及在大样本数据下传统特征降维算法效率普遍不高的现状，提出了结合模拟退火和粒子群算法的混合智能优化算法（SA-PSO）来解决情感特征选择的问题，并结合带权重的离散KNN分类算法（WD-KNN），充分利用情感样本信息进行特征分类。通过对实验仿真数据的分析和与其他方法识别结果的比对，提高了识别率和效率，验证了算法的有效性。

关键词: 情感计算, 情感识别, 特征选择, 混合智能优化

Abstract:

Developing a machine's ability to recognize emotion states is one of the hallmarks of emotional intelligence and important prerequisite for high-level human computer interaction（HCI）.Recording and recognizing physiological signals of emotion has become an increasingly important field of research in affective computing and HCI.For the problem of feature redundancy of physiological signals-based emotion recognition and low efficiency of traditional feature reduction algorithms on great sample data,a hybrid intelligent optimization algorithm based on the simulated annealing algorithm and particle swarm optimization algorithm（SA-PSO）was proposed to solve the problem of emotion feature selection.Then a weighted discrete-KNN classifier（WD-KNN）was presented to classify features by making full use of emotion sample information.The recognition rate and efficiency was increased and the algorithm's validity was verified through the analysis of experimental simulation data and the comparison of several recognition methods.

Key words: affective computing, emotion recognition, feature selection, hybrid intelligent optimization

王海宁,孙守迁,吴剑锋. 基于混合智能优化算法的生理信号情感识别[J]. 电信科学, 2010, 26(9): 129-135.

Haining Wang,Shouqian Sun,Jianfeng Wu. Research on Emotion Recognition with Physiological Signals Based on Hybrid Intelligent Optimization Algorithm[J]. Telecommunications Science, 2010, 26(9): 129-135.

图/表 4

图1

图2

表1

表2

参考文献 16

1	PicardR ． Affective computing． MA: MIT press， 1997
2	HealeyJ . Wearable and automotive systems for affect recognition from physiology. PhD thesis, MIT,Cambridge May2000
3	PicardR , VyzasE , HealeyJ . Toward machine emotional intelligence: analysis of affective physiological state. IEEE Transactions Pattern Analysis and Machine Intelligence, 2001,23（10）: 1175～1191
4	HaagA , GoronzyS , SchaichP ,et al. Emotion recognition using bio-sensors: first step towards an automatic system.In: Affective Dialogue Systems,Tutorial and Research Workshop, Kloster Irsee,Germany, June2004
5	NasozF , AlcarezK , LisettiC ,et al. Emotion recognition from physiological signals for presence technologies. International Journal of Cognition,Technology and Work,Special Issue on Presence, 2003,6（1）
6	ArkW , DryerD , LuD ,et al. The emotion mouse. 8th Int,Conf,Human-computer Interaction, 1999
7	PicardR , KleinJ . Computers that recognize and respond to user emotion: theoretical and practical implications. Interacting Computers, 2002,14（2）: 141～169
8	PudilP , NovovicovaJ , KittlerJ . Floating search methods in feature selection. Pattern Recognition Letters, 1994,15（11）: 1119～1125
9	WagnerJ , KimJ , AndrE . From physiological signals to emotions: Implementing and comparing selected methods for feature extraction and classification.In: Proceedings of the 2005 IEEE International Conference on Multimedia ＆Expo, 2005
10	潘全科，朱剑英．基于进化算法和模拟退火算法的混合调度算法．机械工程学报， 2005,41（6）: 224～227
11	KirkpatrickS , GelattC , VecchiM . Optimization by simulated annealing. Science, 1983,220（4598）: 671～680
12	潘全科，王文宏，朱剑英．基于粒子群优化和模拟退火的混合调度算法．机械工程学报， 2006,17（10）: 1044～1046
13	边肇祺．模式识别．北京：清华大学出版社， 1988
14	PaoTsanglong , ChenYute , LuJhihjheng ,et al. The construction and testing of a mandarin emotional speech database.In: Proceeding of ROCLING XVI, Sep2004
15	杨瑞请．基于BPSO的生理信号的情感状态识别．重庆：西南大学， 2008
16	王玉晶．基于禁忌搜索算法的生理信号情感识别研究．重庆：西南大学， 2008

生理信号	基于混合智能优化和带权重WD-KNN的识别				SFFS识别率	参考文献[15]识别率	参考文献[16]识别率
生理信号	原始特征数	最优子集数	最好识别率	平均识别率	SFFS识别率	参考文献[15]识别率	参考文献[16]识别率
EMG	21	6	84%	81.3%	83%	68%	67.8%
SC	21	5	62%	64.3%	50%	47%	52%
RSP	67	10	80%	75.6%	76%	73%	72.8%
ECG	84	9	75%	71.1%	68%	67%	62.5%
ALL	193	11	93%	92.2%	88.3%	85%	81.5%

情感状态	生理信号	原始特征	最优子集数	最好识别率	最差识别率	平均识别率
高兴	EMG	21	7	81%	72%	75%
	SC	21	6	66%	57%	60%
	RSP	67	28	55%	53%	54%
	ECG	84	32	89%	82%	87%
生气	EMG	21	9	79%	71%	75%
	SC	21	10	61%	53%	56%
	RSP	67	29	99%	95%	97%
	ECG	84	35	52%	50%	51%
悲伤	EMG	21	6	73%	65%	70%
	SC	21	10	51%	35%	45%
	RSP	67	32	69%	64%	68%
	ECG	84	42	78%	69%	75%
喜悦	EMG	21	9	72%	67%	70%
	SC	21	7	50%	34%	38%
	RSP	67	38	41%	35%	37%
	ECG	84	32	40%	31%	35%

[1]	张永, 刘纪奎, 柯文龙. 基于并行可分离卷积和标签平滑正则化的脑电情感识别[J]. 电信科学, 2023, 39(5): 116-128.
[2]	刘国庆, 王兴起, 魏丹, 方景龙, 邵艳利. 基于最大信息系数的软件缺陷数目预测特征选择方法[J]. 电信科学, 2021, 37(5): 133-147.
[3]	林锋,徐柳婧,陈晓华,戚伟强,陈可,朱添田. 一种基于多视角特征融合的Webshell检测方法[J]. 电信科学, 2020, 36(6): 125-132.
[4]	王金华,应娜,朱辰都,刘兆森,蔡哲栋. 基于语谱图提取深度空间注意特征的语音情感识别算法[J]. 电信科学, 2019, 35(7): 100-108.
[5]	章坚武,杨佳佳,吴震东. 一种基于指纹和声纹决策级融合识别方法[J]. 电信科学, 2018, 34(3): 32-40.
[6]	刘洺辛,陈晶,王麒媛. 基于改进特征选择方法的文本情感分类研究[J]. 电信科学, 2018, 34(10): 85-95.
[7]	张春琴,谢立春. 云环境中改进FCM和规则参数优化的网络入侵检测方法[J]. 电信科学, 2018, 34(1): 72-79.
[8]	许小媛,黄黎. 基于实例学习和协同子集搜索的特征选择方法[J]. 电信科学, 2017, 33(6): 105-113.
[9]	陈昊,卿斯汉. 基于组合式算法的Android恶意软件检测方法[J]. 电信科学, 2016, 32(10): 15-21.
[10]	边凌燕,贺仁龙,姚晓辉. 基于DPI数据挖掘实现URL分类挂载的相关技术研究[J]. 电信科学, 2013, 29(11): 6-11.
[11]	陆慧娟,张金伟,张金伟,张金伟,马小平,杨小兵. 基于特征选择的过抽样算法的研究[J]. 电信科学, 2012, 28(1): 91-95.
[12]	陆慧娟,张金伟,马小平,杨小兵. 基于特征选择的过抽样算法的研究[J]. 电信科学, 2012, 28(1): 87-91.