多模态维度情感预测综述

李霞; 卢官明; 闫静杰; 张正言

doi:10.16383/j.aas.2018.c170644

多模态维度情感预测综述

doi: 10.16383/j.aas.2018.c170644

李霞^1,2,,
卢官明^1, ,,
闫静杰^1,,
张正言^1,3,

1.
南京邮电大学通信与信息工程学院南京 210003
2.
安徽工业大学数理学院马鞍山 243000
3.
江苏科技大学电子信息学院镇江 212003

基金项目:

江苏省研究生创新项目 KYLX150827

江苏省自然科学基金 BK20150855

国家自然科学基金 61501249

江苏省重点研发计划项目 BE2016775

江苏省研究生创新项目 KYLX160660

国家自然科学基金 61071167

详细信息

作者简介:
李霞  南京邮电大学通信与信息工程学院博士研究生.2002年获得曲阜师范大学数学与应用数学系学士学位, 2005年获得南京大学应用数学系硕士学位.主要研究方向为情感计算, 模式识别, 机器学习和计算机视觉.E-mail:lx19800102@163.com

闫静杰  南京邮电大学通信与信息工程学院讲师.2006年和2009年获得中国矿业与技术大学电子科学与技术学士学位和信号与信息处理硕士学位.2014年获得东南大学信息与通信工程博士学位.主要研究方向为模式识别, 情感计算, 计算机视觉和机器学习.E-mail:yanjingjie1212@163.com

张正言  南京邮电大学通信与信息工程学院博士研究生.2004年和2007年获得江苏科技大学电子信息工程学士学位和信号与信息处理硕士学位.主要研究方向为模式识别, 机器学习和计算机视觉.E-mail:zhangzhengyan@just.edu.cn

通讯作者:
卢官明南京邮电大学通信与信息工程学院教授.1985年和1988年获得南京邮电大学无线电工程学士学位和通信与电子系统硕士学位, 1999年获得上海交通大学通信与信息系统博士学位.主要研究方向为图像处理, 情感计算, 机器学习.本文通信作者.E-mail:lugm@njupt.edu.cn

计量
- 文章访问数: 4346
- HTML全文浏览量: 1841
- PDF下载量: 1203
- 被引次数: 0
出版历程
- 收稿日期: 2017-11-15
- 录用日期: 2018-03-07
- 刊出日期: 2018-12-20

A Survey of Dimensional Emotion Prediction by Multimodal Cues

LI Xia^{1,2
,},
LU Guan-Ming^{1
, ,},
YAN Jing-Jie^1
,,
ZHANG Zheng-Yan^{1,3
,}

1.
College of Telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications, Nanjing 210003
2.
School of Mathematics and Physics, Anhui University of Technology, Maanshan 243000
3.
School of Electronics and Information, Jiangsu University of Science and Technology, Zhenjiang 212003

Funds:

Jiangsu Innovation Program for Graduate Education KYLX150827

Natural Science Foundation of Jiangsu Province BK20150855

National Natural Science Foundation of China 61501249

Key Research and Development Program of Jiangsu Province BE2016775

Jiangsu Innovation Program for Graduate Education KYLX160660

National Natural Science Foundation of China 61071167

More Information

Author Bio:
Ph. D. candidate at the College of Telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications. She received her bachelor degree in mathematics and applied mathematics from Qufu Normal University in 2002 and master degree in applied mathematics from Nanjing University in 2005, respectively. Her research interest covers affective computing, pattern recognition, machine learning, and computer vision

Lecturer at the College of Telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications. He received his bachelor degree in electronic science and technology in 2006 and master degree in signal and information processing in 2009 from China University of Mining and Technology, and Ph. D. degree in signal and information processing from Southeast University in 2014. His research interest covers pattern recognition, affective computing, computer vision, and machine learning

Ph. D. candidate at the College of Telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications. He received his bachelor degree in electronic information engineering and master degree in signal and information processing from Jiangsu University of Science and Technology in 2004 and 2007, respectively. His research interest covers pattern recognition, machine learning, and computer vision

Corresponding author: LU Guan-Ming Professor at the College of Communication and Information Engineering, Nanjing University of Posts and Telecommunications. He received his bachelor degree in radio engineering and master degree in communication and electronic systems from Nanjing University of Posts and Telecommunications in 1985 and 1988, respectively, and Ph. D. degree in communication and information systems from Shanghai Jiao Tong University in 1999. His research interest covers image processing, affective computing, and machine learning. Corresponding author of this paper

摘要

摘要: 维度情感模型通过几个取值连续的维度（如唤醒维、效价维、支配维等）将情感刻画为一个多维信号.与传统的离散情感模型相比，具有表示情感的范围广、能描述情感的演变过程等优点，近年来受到越来越多情感识别研究者的关注.多模态维度情感预测是一项复杂的工程，预测性能受所使用的模态、每个模态的特征提取、信息融合技术、标注人员的标注误差等多方面影响.为了提高多模态维度情感预测的性能，研究者在各个方面都做出了不懈努力.本文综述了维度情感的概念、标注，维度情感预测的性能评价指标以及多模态维度情感预测的研究现状，对比和分析了各种因素对多模态维度情感预测性能的影响，并总结出多模态维度情感预测面临的挑战及发展趋势.
- 情感识别 /
- 情感预测 /
- 维度情感模型 /
- 离散情感模型 /
- 信息融合 /
- 特征提取
Abstract: The dimensional emotion model characterizes emotion as a signal in a multi-dimensional space spanned by several continuously valued dimensions (such as arousal, valence, and dominance). Compared with the discrete emotion model, it has the advantages that it can distinguish subtle difference of emotion, can represent evolution of emotion, etc. So the dimensional emotion model has been paid more and more attention in recent years. Dimensional emotion prediction from multimodal cues is a complex task, the prediction performance is influenced by such as modalities used, features extracted from each modality, information fusion technique, annotation errors. In order to improve multimodal dimensional emotion prediction performance, researchers have made persistent efforts in all aspects. In the paper, concept and annotation of dimensional emotion, performance evaluation criteria of dimensional emotion prediction, and research status of multimodal dimensional emotion prediction are reviewed; influences of various factors on emotion prediction performance are analyzed; challenge and development trend of multimodal dimensional emotion prediction are summarized.
- Emotion recognition /
- emotion prediction /
- dimensional emotion model /
- discrete emotion model /
- information fusion /
- feature extraction
注释:

1) 本文责任编委黄庆明

HTML全文

图 1 愉悦-唤醒模型

Fig. 1 Pleasure-arousal model

下载: 全尺寸图片幻灯片

图 2 SAM系统

Fig. 2 SAM system

下载: 全尺寸图片幻灯片

图 3 FEELtrace标注示例

Fig. 3 Example of FEELtrace annotation

下载: 全尺寸图片幻灯片

图 4 ANNEMO标注示例

Fig. 4 Example of ANNEMO annotation

下载: 全尺寸图片幻灯片

图 5 具有不同MSE和CC的效价维的预测与真值的对比图

Fig. 5 Comparison of the prediction and truth values of valence dimension with different MSEs and CCs

下载: 全尺寸图片幻灯片

图 6 CC相同的条件下唤醒维的预测与真值的对比图

Fig. 6 Comparison of the prediction and truth values of arousal dimension with the same CC

下载: 全尺寸图片幻灯片

图 7 单层卷积神经网络的三个阶段

Fig. 7 The tree phases of the single layer convolutional neural network

下载: 全尺寸图片幻灯片

图 8 自编码器的结构

Fig. 8 Structure of autoencoder

下载: 全尺寸图片幻灯片

图 9 RNN的网络结构

Fig. 9 Network structure of RNN

下载: 全尺寸图片幻灯片

图 10 OA融合框架

Fig. 10 OA fusion framework

下载: 全尺寸图片幻灯片

表 1 常用维度情感数据库总结

Table 1 Summary of the frequently used dimensional emotion

数据库	场景	参与者数	模态	情感维度	标注者数	工具/方法	标签范围与类型
SEMAINE	Solid SAL	24	Vi + Au	A, V, E, D, I	2~8人	FEELtrace	[-1, 1]的连续值
RECOLA	远程视频会议	46	Vi + Au + Ph	A, V	6人	ANNEMO	[-1, 1]的连续值
IEMOCAP	双人对话表演	10	Vi + Au	A, V, D	至少2人	SAM系统	1~5的整数值
CreativeIT	双人对话表演	16	Vi + Au	A, V, D	3~4人	FEELtrace	[-1, 1]的连续值
DEAP	观看音乐视频	32	Vi + Ph	A, V, D	1人	SAM系统	[1, 9]的连续值
VAM	电视脱口秀	47	Vi + Au	A, V, D	6~34人	SAM系统	[-1, 1]的5点等间隔值
注: Vi —视觉模态, Au —听觉模态, Ph —生理信号, A —唤醒维, V —效价维, E —期望维, D —支配维, I —强度维

下载: 导出CSV

表 2 维度情感预测文献总结

Table 2 Literature review of the dimensional emotion

文献 (出版日期)	模态		特征	特征选择和降维	维度情感预测模型		信息融合方法
文献 (出版日期)	模态		特征	特征选择和降维	回归模型	分类模型	信息融合方法
[49] (2008)	Au		声学特征	CFS	LSTM-RNN	CRF	-
[16] (2009)	Au		声学特征	-	-	HMM	-
[28] (2010)	Vi	头部运动	几何特征	-	SVR	-	-
[50] (2010)	Vi	步态	几何特征	PCA, KPCA, LDA, GDA	-	NN	-
[18] (2010)	Au	声音	声学特征	CFS	-	LSTM-RNN	FE
[18] (2010)	Au	语言	语言特征	CFS	-	LSTM-RNN	FE
[51] (2010)	Au	声音	声学特征	-	LSTM-RNN	-	FE
[51] (2010)	Au	语言	语言特征	-	LSTM-RNN	-	FE
[52] (2010)	Vi		几何特征	PCA, CFS	-	BLSTM	FE
[52] (2010)	Au		声学特征	PCA, CFS	-	BLSTM	FE
[48] (2011)	Vi		LBP特征	PCA, SPCA	SVR	-	FE + DE
[48] (2011)	Au		声学特征	PCA, SPCA	SVR	-	FE + DE
[53] (2011)	Vi		几何特征	-	GMM	-	MO
[53] (2011)	Au		声学特征	-	GMM	-	MO
[32] (2011)	Vi		头部姿势、面部运动单元	CFS	SVR	-	FE-基于串的方法
[32] (2011)	Au		笑声、叹息声、关键词	CFS	SVR	-	FE-基于串的方法
[45] (2011)	Au	声音	声学特征	CFS	SVR	-	FE
[45] (2011)	Au	语言	BoCNG特征	CFS	SVR	-	FE
[21] (2011)	Vi	面部	几何特征	-	BLSTM	-	OA
	Vi	肩膀	几何特征	-	BLSTM	-	OA
		Au	声学特征
[54] (2012)	Vi		几何特征	-	-	EWSC-HMM	MO
[54] (2012)	Au		声学特征	-	-	EWSC-HMM	MO
[55] (2012)	Vi	面部	面部表情			-
	Vi	身体	几何特征	-	多模态推断系统		MO
	Au		语言与关键词信息
[56] (2012)	Vi		多尺度动态视频特征	新的基于相关的特征选择	核回归	-	OA-局部线性回归
[56] (2012)	Au		声学特征	新的基于相关的特征选择	核回归	-	OA-局部线性回归
[57] (2012)	Vi	面部	几何特征	-	OA-RVM	-	OA-RVM
	Vi	肩膀	几何特征	-	OA-RVM	-	OA-RVM
	Au		声学特征
[31] (2013)	Vi		基于光流的低级特征
	Au	声音	声学特征	CFS	BLSTM	-	FE
	Au	语言	BoW特征
[58] (2013)	Vi		局部时空特征	-	SVR	-	DE-加权和
[58] (2013)	Au		声学特征	-	SVR	-	DE-加权和
[59] (2013)	Vi		几何特征	CSR	CSR	-	CSR
[59] (2013)	Au		声学特征	CSR	CSR	-	CSR
[60] (2015)	Vi		EOH, LBP, LBQ	-	PLS	-	DE-线性回归
[60] (2015)	Au		声学特征	-	PLS	-	DE-线性回归
[23] (2015)	Vi		LBP-TOP, LGBP-TOP, PHOG-TOP, HOG, 时空几何特征	-	随机森林	-	DE-平均
	Au		声学特征
	Ph		生理特征
[61] (2015)	Vi		LGBP-TOP, 时空几何特征	-	SVM, RVM	-	OA
	Au		声学特征
	Ph		时间和频域特征
[62] (2015)	Vi		LGBP-TOP, LPQ-TOP, 时空几何特征	-	DBLSTM	-	DE-DBLSTM
	Au		声学特征
	Ph		时间和频域特征
[63] (2015)	Vi		LGBP-TOP, 时空几何特征	PCA	LSTM	-	FE, DE-线性回归
	Au		声学特征
	Ph		时间和频域特征
[64] (2016)	Au		声学特征	-	DBLSTM	-	DE-ELM
[65] (2016)	Au		加强后的声学特征	-	SVR	-	-
[66] (2016)	Vi		LBP特征	CFS	DNN-SKF	-	FE
	Au		声学特征
	词汇		词汇特征
[67] (2016)	Vi		CNN特征	-	LSTM	-	DE-Kalman滤波
	Au		声学特征
	Ph		时间和频域特征
[68] (2016)	Au		CNN特征	-	LSTM	-	-
[69] (2016)	Vi		LGBP-TOP, 几何特征, CNN特征	PCA	LSTM	-	DE-LSTM
	Au		声学特征
	Ph		时间和频域特征
[20] (2017)	Ph		通过SAE进行抽象的传统特征	-	Bayesian模型	-	FE-分层的特征融合网络
注: 若文中使用多种方法进行对比分析, 这里只列出性能最好的一种方法. Vi — 视觉模态, Au — 听觉模态, Ph — 生理信号, FE — 特征层融合, DE — 决定层融合 (决定层融合使用的具体方法), MO — 模型层融合, OA — 输出相关融合

下载: 导出CSV

表 3 连续维度情感预测对比总结

Table 3 Comparison and summary of continuous dimensional emotion prediction

文献	数据库	模态	情感维度		特征	回归模型	融合方法	延时弥补 (Y/N)	延时弥补 (Y/N)	最好平均预测性能
文献	数据库	模态	情感维度		特征	回归模型	融合方法	延时弥补 (Y/N)	延时弥补 (Y/N)	CC	CCC
[78] (基准)	AVEC 2012	Vi	A, V, E, D	Vi	LBP	SVR	-	N	N	0.09	-
[78] (基准)	AVEC 2012	Vi+Au	A, V, E, D	Au	声学特征	SVR	FE	N	N	0.11	-
[58]	AVEC 2012	Vi	A, V, E, D	Vi	局部时空特征	SVR	-	N	N	0.41	-
[58]	AVEC 2012	Vi+Au	A, V, E, D	Au	声学特征	SVR	DE-加权和	N	N	0.42	-
[79] (基准)	AVEC 2014	Vi	A, V, D	Vi	LGBP-TOP	SVR	-	N	N	0.20	-
[79] (基准)	AVEC 2014	Vi+Au	A, V, D	Au	声学特征	SVR	DE-加权和	N	N	0.36	-
[22] (基准)	AVEC 2015	Vi	A, V	Vi	LGBP-TOP+时空几何特征	SVR	-	N	N	0.29	0.20
		Vi+Au+Ph		Au	声学特征		DE-线性回归	N	N	0.42	0.41
		Vi+Au+Ph		Ph	时间和频域特性
[47] (基准)	AVEC 2016	Vi	A, V	Vi	LGBP-TOP+时空几何特征	SVR	-	N	N	-	0.40
		Vi+Au+Ph		Au	声学特征		DE-线性回归	Y	N	-	0.66
		Vi+Au+Ph		Ph	时间和频域特性
[55]	AVEC 2012	Vi+Au	A, V, E, D	Vi	面部表情+身体语言	多模态模糊推断系统	MO	N	N	0.43	-
[55]	AVEC 2012	Vi+Au	A, V, E, D	Au	语句和关键词信息	多模态模糊推断系统
[56]	AVEC 2012	Vi+Au	A, V, E, D	Vi	多尺度动态视频特征	核回归	OA-局部线性回归	Y	Y	0.46	-
[56]	AVEC 2012	Vi+Au	A, V, E, D	Au	声学特征	核回归	OA-局部线性回归	Y	Y	0.46	-
[61]	AVEC 2015	Vi+Au+Ph	A, V	Vi	LGBP-TOP+时空几何特征	SVM, RVM	OA-Regression	Y	Y	-	0.66
	AVEC 2015	Vi+Au+Ph	A, V	Au	声学特征
				Ph	时间和频域特性
[62]	AVEC 2015	Vi+Au+Ph	A, V	Vi	LGBP-TOP+LPQ-TOP+时空几何特征	DBLSTM	DE-DBLSTM	Y	N	0.68	0.68
	AVEC 2015	Vi+Au+Ph	A, V	Au	声学特征
				Ph	时间和频域特性
注: Vi —视觉模态, Au —听觉模态, Ph —生理信号, A —唤醒维, V —效价维, E —期望维, D —支配维, FE —特征层融合, DE —决定层融合(决定层融合使用的具体方法), MO —模型层融合, OA —输出相关融合

下载: 导出CSV

表 4 维度情感分类对比总结

Table 4 Comparison and summary of dimensional emotion

文献	数据库	模态	情感维度	特征		识别模型	信息融合方法	最好平均性能(%)
文献	数据库	模态	情感维度	特征		识别模型	信息融合方法	WA	UA
[83] (基准)	AVEC 2011	音频	A, V, E, D	声学特征		SVM	-	45.05	51.95
[31]	AVEC 2011	音频	A, V, E, D	声学特征		LSTM	-	65.2	58.5
[54]	SEMAINE	音频+视频	A, V	视频	几何特征	EWSC-HMM	模型层融合	-	78.1
[54]	SEMAINE	音频+视频	A, V	音频	声学特征	EWSC-HMM	模型层融合	-	78.1
[80]	SEMAINE	音频+视频	A, V	视频	几何特征	2H-SC-HMM	模型层融合	-	87.5
[80]	SEMAINE	音频+视频	A, V	音频	声学特征	2H-SC-HMM	模型层融合	-	87.5
注: A —唤醒维, V —效价维, E —期望维, D —支配维, UA —未加权准确性, WA —加权准确性

下载: 导出CSV

参考文献(83)

[1]	刘烨, 付秋芳, 傅小兰.认知与情绪的交互作用.科学通报, 2009, 54(18):2783-2796 http://d.old.wanfangdata.com.cn/Periodical/xlkx200603052 Liu Ye, Fu Qiu-Fang, Fu Xiao-Lan. The interaction between cognition and emotion. Chinese Science Bulletin, 2009, 54(22):4102-4116 http://d.old.wanfangdata.com.cn/Periodical/xlkx200603052
[2]	D'Mello S K, Kory J. A review and meta-analysis of multimodal affect detection systems. ACM Computing Surveys, 2015, 47(3):Article No. 43 http://dl.acm.org/citation.cfm?id=2682899
[3]	Zeng Z H, Pantic M, Roisman G I, Huang T S. A survey of affect recognition methods:audio, visual, and spontaneous expressions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(1):39-58 doi: 10.1109/TPAMI.2008.52
[4]	Poria S, Cambria E, Bajpai R, Hussain A. A review of affective computing:from unimodal analysis to multimodal fusion. Information Fusion, 2017, 37:98-125 doi: 10.1016/j.inffus.2017.02.003
[5]	乐国安, 董颖红.情绪的基本结构:争论、应用及其前瞻.南开学报(哲学社会科学版), 2013, (1):140-150 http://d.old.wanfangdata.com.cn/Periodical/yejy-jykx201305009 Yue Guo-An, Dong Ying-Hong. On the categorical and dimensional approaches of the theories of the basic structure of emotions. Nankai Journal (Literature and Social Science Edition), 2013, (1):140-150 http://d.old.wanfangdata.com.cn/Periodical/yejy-jykx201305009
[6]	Arifin S, Cheung P Y K. Affective level video segmentation by utilizing the pleasure-arousal-dominance information. IEEE Transactions on Multimedia, 2008, 10(7):1325-1341 doi: 10.1109/TMM.2008.2004911
[7]	Cowie R, Douglas-Cowie E, Savvidou S, McMahon E, Sawey M, Schröder M. "FEELTRACE": an instrument for recording perceived emotion in real time. In: Proceedings of the 2000 ISCA Tutorial and Research Workshop on Speech and Emotion. Northern Ireland: ISCA, 2000. 19-24
[8]	韩文静, 李海峰, 阮华斌, 马琳.语音情感识别研究进展综述.软件学报, 2014, 25(1):37-50 http://d.old.wanfangdata.com.cn/Periodical/rjxb201401004 Han Wen-Jing, Li Hai-Feng, Ruan Hua-Bin, Ma Lin. Review on speech emotion recognition. Journal of Software, 2014, 25(1):37-50 http://d.old.wanfangdata.com.cn/Periodical/rjxb201401004
[9]	Gunes H, Schuller B. Categorical and dimensional affect analysis in continuous input:current trends and future directions. Image and Vision Computing, 2013, 31(2):120-136 doi: 10.1016/j.imavis.2012.06.016
[10]	Fontaine J R J, Scherer K R, Roesch E B, Eiisworth P C. The world of emotions is not two-dimensional. Psychological Science, 2007, 18(12):1050-1057 doi: 10.1111/j.1467-9280.2007.02024.x
[11]	邹吉林, 张小聪, 张环, 于靓, 周仁来.超越效价和唤醒-情绪的动机维度模型述评.心理科学进展, 2011, 19(9):1339-1346 http://d.old.wanfangdata.com.cn/Conference/8110806 Zou Ji-Lin, Zhang Xiao-Cong, Zhang Huan, Yu Liang, Zhou Ren-Lai. Beyond dichotomy of valence and arousal:review of the motivational dimensional model of affect. Advances in Psychological Science, 2011, 19(9):1339-1346 http://d.old.wanfangdata.com.cn/Conference/8110806
[12]	Morris J D. Observations:SAM:the self-assessment manikin-an efficient cross-cultural measurement of emotional response. Journal of Advertising Research, 1995, 35:63-68 http://d.old.wanfangdata.com.cn/Periodical/kjkxxb201702001
[13]	Koelstra S, Muhl C, Soleymani M, Lee J S, Yazdani A, Ebrahimi T, et al. DEAP:a database for emotion analysis using physiological signals. IEEE Transactions on Affective Computing, 2012, 3(1):18-31 doi: 10.1109/T-AFFC.2011.15
[14]	Busso C, Bulut M, Lee C C, Kazemzadeh A, Mower E, Kim S, et al. IEMOCAP:interactive emotional dyadic motion capture database. Language Resources and Evaluation, 2008, 42(4):335-359 doi: 10.1007/s10579-008-9076-6
[15]	Ringeval F, Sonderegger A, Sauer J, Lalanne D. Introducing the RECOLA multimodal corpus of remote collaborative and affective interactions. In: Proceedings of the 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition. Shanghai, China: IEEE, 2013. 1-8 http://www.mendeley.com/catalog/introducing-recola-multimodal-corpus-remote-collaborative-affective-interactions/
[16]	Schuller B, Vlasenko B, Eyben F, Rigoll G, Wendemuth A. Acoustic emotion recognition: a benchmark comparison of performances. In: Proceedings of the 2009 IEEE Workshop on Automatic Speech Recognition and Understanding. Merano, Italy: IEEE, 2009. 552-557 http://www.mendeley.com/catalog/acoustic-emotion-recognition-benchmark-comparison-performances/
[17]	Tarasov A, Delany S J. Benchmarking classification models for emotion recognition in natural speech: a multi-corporal study. In: Proceedings of the 2011 IEEE International Conference on Automatic Face and Gesture Recognition and Workshops. Santa Barbara, CA, USA: IEEE, 2011. 841-846 http://www.mendeley.com/catalog/benchmarking-classification-models-emotion-recognition-natural-speech-multicorporal-study/
[18]	Wöllmer M, Schuller B, Eyben F, Rigoll G. Combining long short-term memory and dynamic Bayesian networks for incremental emotion-sensitive artificial listening. IEEE Journal of Selected Topics in Signal Processing, 2010, 4(5):867-881 doi: 10.1109/JSTSP.2010.2057200
[19]	Espinosa H P, García C A R, Pineda L V. Features selection for primitives estimation on emotional speech. In: Proceedings of the 2010 IEEE International Conference on Acoustics Speech and Signal Processing. Dallas, TX, USA: IEEE, 2010. 5138-5141 http://www.mendeley.com/research/features-selection-primitives-estimation-emotional-speech/
[20]	Yin Z, Zhao M Y, Wang Y X, Yang J D, Zhang J H. Recognition of emotions using multimodal physiological signals and an ensemble deep learning model. Computer Methods and Programs in Biomedicine, 2017, 140:93-110 doi: 10.1016/j.cmpb.2016.12.005
[21]	Nicolaou M A, Gunes H, Pantic M. Continuous prediction of spontaneous affect from multiple cues and modalities in valence-arousal space. IEEE Transactions on Affective Computing, 2011, 2(2):92-105 doi: 10.1109/T-AFFC.2011.9
[22]	Ringeval F, Schuller B, Valstar M, Jaiswal S, Marchi E, Lalanne D, et al. AV+EC 2015: the first affect recognition challenge bridging across audio, video, and physiological data. In: Proceedings of the 5th International Workshop on Audio/Visual Emotion Challenge. Brisbane, Australia: ACM, 2015. 3-8 doi: 10.1145/2808196.2811642
[23]	Kächele M, Schels M, Thiam P, Schwenker F. Fusion mappings for multimodal affect recognition. In: Proceedings of the 2015 IEEE Symposium Series on Computational Intelligence. Cape Town, South Africa: IEEE, 2015. 307-313 http://www.deepdyve.com/lp/institute-of-electrical-and-electronics-engineers/fusion-mappings-for-multimodal-affect-recognition-oJNU0tN0pn
[24]	孙晓, 潘汀, 任福继.基于ROI-KNN卷积神经网络的面部表情识别.自动化学报, 2016, 42(6):883-891 http://www.aas.net.cn/CN/abstract/abstract18879.shtml Sun Xiao, Pan Ting, Ren Fu-Ji. Facial expression recognition using ROI-KNN deep convolutional neural networks. Acta Automatica Sinica, 2016, 42(6):883-891 http://www.aas.net.cn/CN/abstract/abstract18879.shtml
[25]	徐峰, 张军平.人脸微表情识别综述.自动化学报, 2017, 43(3):333-348 http://www.aas.net.cn/CN/abstract/abstract19013.shtml Xu Feng, Zhang Jun-Ping. Facial microexpression recognition:a survey. Acta Automatica Sinica, 2017, 43(3):333-348 http://www.aas.net.cn/CN/abstract/abstract19013.shtml
[26]	Ekman P. Universal facial expressions of emotion. California Mental Health Research Digest, 1970, 8(4):151-158 http://d.old.wanfangdata.com.cn/OAPaper/oai_pubmedcentral.nih.gov_3358835
[27]	Kleinsmith A, Bianchi-Berthouze N. Affective body expression perception and recognition:a survey. IEEE Transactions on Affective Computing, 2013, 4(1):15-33 doi: 10.1109/T-AFFC.2012.16
[28]	Gunes H, Pantic M. Dimensional emotion prediction from spontaneous head gestures for interaction with sensitive artificial listeners. In: Proceeding of the 10th International Conference on Intelligent Virtual Agents. Berlin, Heidelberg, Germany: Springer-Verlag, 2010. 371-377 doi: 10.1007%2F978-3-642-15892-6_39
[29]	Metallinou A, Yang Z J, Lee C C, Busso C, Carnicke S, Narayanan S. The USC CreativeIT database of multimodal dyadic interactions:from speech and full body motion capture to continuous emotional annotations. Language Resources and Evaluation, 2016, 50(3):497-521 doi: 10.1007/s10579-015-9300-0
[30]	王科, 夏睿.情感词典自动构建方法综述.自动化学报, 2016, 42(4):495-511 http://www.aas.net.cn/CN/abstract/abstract18838.shtml Wang Ke, Xia Rui. A survey on automatical construction methods of sentiment lexicons. Acta Automatica Sinica, 2016, 42(4):495-511 http://www.aas.net.cn/CN/abstract/abstract18838.shtml
[31]	Wöllmer M, Kaiser M, Eyben F, Schuller B, Rigoll G. LSTM-Modeling of continuous emotions in an audiovisual affect recognition framework. Image and Vision Computing, 2013, 31(2):153-163 http://dl.acm.org/citation.cfm?id=2438270
[32]	Eyben F, Wöllmer M, Valstar M F, Gunes H, Schuller B, Pantic M. String-based audiovisual fusion of behavioural events for the assessment of dimensional affect. In: Proceedings of the 2011 IEEE International Conference on Automatic Face and Gesture Recognition and Workshops. Santa Barbara, CA, USA: IEEE, 2011. 322-329 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.364.3773
[33]	彭聃龄.普通心理学.北京:北京师范大学出版社, 2001. Peng Ran-Ling. General Psychology. Beijing:Beijing Normal University Press, 2001.
[34]	Calvo R A, D'Mello S. Affect detection:an interdisciplinary review of models, methods, and their applications. IEEE Transactions on Affective Computing, 2010, 1(1):18-37 doi: 10.1109/T-AFFC.2010.1
[35]	Mckeown G, Valstar M, Cowie R, Pantic M, Schroder M. The SEMAINE database:annotated multimodal records of emotionally colored conversations between a person and a limited agent. IEEE Transactions on Affective Computing, 2012, 3(1):5-17 doi: 10.1109/T-AFFC.2011.20
[36]	Grimm M, Kroschel K, Narayanan S. The Vera am Mittag German audio-visual emotional speech database. In: Proceedings of the 2008 IEEE International Conference on Multimedia and Expo. Hannover, German: IEEE, 2008. 865-868 http://www.mendeley.com/catalog/vera-mittag-german-audiovisual-emotional-speech-database/
[37]	Lades M, Vorbruggen J C, Buhmann J, Lang J, von der Malsburg C, Wurtz R P, et al. Distortion invariant object recognition in the dynamic link architecture. IEEE Transactions on Computers, 1993, 42(3):300-311 doi: 10.1109/12.210173
[38]	Ahonen T, Hadid A, Pietikainen M. Face description with local binary patterns:application to face recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(12):2037-2041 doi: 10.1109/TPAMI.2006.244
[39]	Dalal N, Triggs B. Histograms of oriented gradients for human detection. In: Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. San Diego, USA: IEEE, 2005. 886-893 http://www.mendeley.com/catalog/histogram-oriented-gradients-human-detection/
[40]	Viola P, Jones M. Rapid object detection using a boosted cascade of simple features. In: Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Kauai, USA: IEEE, 2001. I-511-I-518 http://www.mendeley.com/research/colonialism-homosexuality-review/
[41]	Zhao G Y, Pietikäinen M. Dynamic texture recognition using local binary patterns with an application to facial expressions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(6):915-28 doi: 10.1109/TPAMI.2007.1110
[42]	Jiang B H, Valstar M, Martinez B, Pantic M. A dynamic appearance descriptor approach to facial actions temporal modeling. IEEE Transactions on Cybernetics, 2014, 44(2):161-174 doi: 10.1109/TCYB.2013.2249063
[43]	Almaev T R, Valstar M F. Local Gabor binary patterns from three orthogonal planes for automatic facial expression recognition. In: Proceedings of the 2013 Humaine Association Conference on Affective Computing and Intelligent Interaction. Geneva, Switzerland: IEEE, 2013. 356-361 http://www.mendeley.com/research/local-gabor-binary-patterns-three-orthogonal-planes-automatic-facial-expression-recognition/
[44]	Yang P, Liu Q, Metaxas D N. Boosting coded dynamic features for facial action units and facial expression recognition. In: Proceedings of the 2007 IEEE Conference on Computer Vision and Pattern Recognition. Minneapolis, USA: IEEE, 2007. 1-6 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.581.9566
[45]	Schuller B. Recognizing affect from linguistic information in 3D continuous space. IEEE Transactions on Affective Computing, 2011, 2(4):192-205 doi: 10.1109/T-AFFC.2011.17
[46]	Jenke R, Peer A, Buss M. Feature extraction and selection for emotion recognition from EEG. IEEE Transactions on Affective Computing, 2014, 5(3):327-339 doi: 10.1109/TAFFC.2014.2339834
[47]	Valstar M, Gratch J, Schuller B, Ringeval F, Lalanne D, Torres M T, et al. AVEC 2016: depression, mood, and emotion recognition workshop and challenge. In: Proceedings of the 6th International Workshop on Audio/Visual Emotion Challenge. Amsterdam, The Netherlands: ACM, 2016. 3-10 http://www.deepdyve.com/lp/association-for-computing-machinery/avec-2016-depression-mood-and-emotion-recognition-workshop-and-bJsOVdX1yf
[48]	Sayedelahl A, Araujo R, Kamel M S. Audio-visual feature-decision level fusion for spontaneous emotion estimation in speech conversations. In: Proceedings of the 2013 IEEE International Conference on Multimedia and Expo Workshops. San Jose, CA, USA: IEEE, 2013. 1-6 http://www.deepdyve.com/lp/institute-of-electrical-and-electronics-engineers/audio-visual-feature-decision-level-fusion-for-spontaneous-emotion-qFjpwOTK0Y
[49]	Wöllmer M, Eyben F, Reiter S, Schuller B, Cox C, Douglas-Cowie E, et al. Abandoning emotion classes-towards continuous emotion recognition with modelling of long-range dependencies. In: Proceedings of the 2008 Interspeech. Brisbane, Australia: DBLP, 2008. 597-600
[50]	Karg M, Kuhnlenz K, Buss M. Recognition of affect based on gait patterns. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 2010, 40(4):1050-1061 doi: 10.1109/TSMCB.2010.2044040
[51]	Eyben F, Wöllmer M, Graves A, Schuller B, Douglas-Cowie E, Cowie R. On-line emotion recognition in a 3-D activation-valence-time continuum using acoustic and linguistic cues. Journal on Multimodal User Interfaces, 2010, 3(1-2):7-19 doi: 10.1007/s12193-009-0032-6
[52]	Wöllmer M, Metallinou A, Eyben F, Schuller B, Narayanan S. Context-sensitive multimodal emotion recognition from speech and facial expression using bidirectional LSTM modeling. In: Proceedings of the 11th Annual Conference of the International Speech Communication Association. Makuhari, Chiba, Japan: DBLP, 2010. 2362-2365
[53]	Metallinou A, Katsamanis A, Wang Y, Narayanan S. Tracking changes in continuous emotion states using body language and prosodic cues. In: Proceedings of the 2011 IEEE International Conference on Acoustics, Speech, and Signal Processing. Prague, Czech: IEEE, 2011. 2288-2291 http://www.mendeley.com/catalog/tracking-changes-continuous-emotion-states-using-body-language-prosodic-cues/
[54]	Lin J C, Wu C H, Wei W L. Error weighted semi-coupled hidden Markov model for audio-visual emotion recognition. IEEE Transactions on Multimedia, 2012, 14(1):142-156 doi: 10.1109/TMM.2011.2171334
[55]	Soladié C, Salam H, Pelachaud C, Stoiber N, Séguier R. A multimodal fuzzy inference system using a continuous facial expression representation for emotion detection. In: Proceedings of the 14th ACM International Conference on Multimodal Interaction. Santa Monica, California, USA: ACM, 2012. 493-500 http://www.mendeley.com/research/multimodal-fuzzy-inference-system-using-continuous-facial-expression-representation-emotion-detectio/
[56]	Nicolle J, Rapp V, Bailly K, Prevost L, Chetouani M. Robust continuous prediction of human emotions using multiscale dynamic cues. In: Proceedings of the 14th ACM International Conference on Multimodal Interaction. Santa Monica, California, USA: ACM, 2012: 501-508 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.433.2139
[57]	Nicolaou M A, Gunes H, Pantic M. Output-associative RVM regression for dimensional and continuous emotion prediction. In: Proceedings of the 2012 IEEE International Conference on Automatic Face and Gesture Recognition and Workshops. Santa Barbara, CA, USA: IEEE, 2012. 16-23 http://www.mendeley.com/catalog/outputassociative-rvm-regression-dimensional-continuous-emotion-prediction/
[58]	Song Y, Morency L P, Davis R. Learning a sparse codebook of facial and body microexpressions for emotion recognition. In: Proceedings of the 15th ACM on International Conference on Multimodal Interaction. Sydney, Australia: ACM, 2013. 237-244 http://www.mendeley.com/research/learning-sparse-codebook-facial-body-microexpressions-emotion-recognition/
[59]	Nicolaou M A, Zafeiriou S, Pantic M. Correlated-spaces regression for learning continuous emotion dimensions. In: Proceedings of the 21st ACM International Conference on Multimedia. Barcelona, Spain: ACM, 2013. 773-776 http://www.mendeley.com/research/correlatedspaces-regression-learning-continuous-emotion-dimensions/
[60]	Gaus Y F A, Meng H Y, Jan A, Zhang F, Turabzadeh S. Automatic affective dimension recognition from naturalistic facial expressions based on wavelet filtering and PLS regression. In: Proceedings of the 11th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition. Ljubljana, Yugoslavia: IEEE, 2015. 1-6 http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=7284859
[61]	Huang Z, Dang T, Cummins N, Stasak B, Le P, Sethu V, et al. An investigation of annotation delay compensation and output-associative fusion for multimodal continuous emotion prediction. In: Proceedings of the 2015 International Workshop on Audio/Visual Emotion Challenge. New York, USA: ACM, 2015. 41-48 doi: 10.1145/2808196.2811640
[62]	He L, Jiang D M, Yang L, Pei E C, Wu P, Sahli H. Multimodal affective dimension prediction using deep bidirectional long short-term memory recurrent neural networks. In: Proceedings of the 5th International Workshop on Audio/Visual Emotion Challenge. Brisbane, Australia: ACM, 2015. 73-80
[63]	Chen S Z, Jin Q. Multi-modal dimensional emotion recognition using recurrent neural network. In: Proceedings of the 5th International Workshop on Audio/Visual Emotion Challenge. Brisbane, Australia: ACM, 2015. 49-56 http://dl.acm.org/citation.cfm?id=2811638
[64]	Li X X, Xianyu H, Tian J S, Chen W X, Meng F H, Xu M X, et al. A deep bidirectional long short-term memory based multi-scale approach for music dynamic emotion prediction. In: Proceedings of the 2016 IEEE International Conference on Acoustics, Speech, and Signal Processing. Shanghai, China: IEEE, 2016. 544-548 http://ieeexplore.ieee.org/document/7471734/
[65]	Zhang Z X, Ringeval F, Han J, Deng J, Marchi E, Schuller B. Facing realism in spontaneous emotion recognition from speech: feature enhancement by autoencoder with LSTM neural networks. In: Proceedings of the 2016 Conference of the International Speech Communication Association. San Francisco, USA: ISCA, 2016. 3593-3597
[66]	Pei E C, Xia X H, Yang L, Jiang D M, Sahli H. Deep neural network and switching Kalman filter based continuous affect recognition. In: Proceedings of the 2016 IEEE International Conference on Multimedia and Expo Workshops. Seattle, WA, USA: IEEE, 2016. 1-6 http://www.deepdyve.com/lp/institute-of-electrical-and-electronics-engineers/deep-neural-network-and-switching-kalman-filter-based-continuous-Gi0nak0reF
[67]	Brady K, Gwon Y, Khorrami P, Godoy E, Campbell W, Dagli C, et al. Multi-modal audio, video and physiological sensor learning for continuous emotion prediction. In: Proceedings of the 6th International Workshop on Audio/Visual Emotion Challenge. Amsterdam, The Netherlands: ACM, 2016. 97-104 http://experts.illinois.edu/en/publications/multi-modal-audio-video-and-physiological-sensor-learning-for-con
[68]	Trigeorgis G, Ringeval F, Brueckner R, Marchi E, Nicolaou M A, Schuller B, et al. Adieu features? End-to-end speech emotion recognition using a deep convolutional recurrent network. In: Proceedings of the 2016 IEEE International Conference on Acoustics, Speech and Signal Processing. Shanghai, China: IEEE, 2016. 5200-5204 http://ieeexplore.ieee.org/document/7472669/
[69]	Chao L L, Tao J H, Yang M H, Li Y, Wen Z Q. Long short term memory recurrent neural network based multimodal dimensional emotion recognition. In: Proceedings of the 5th International Workshop on Audio/Visual Emotion Challenge. Brisbane, Australia: ACM, 2015. 65-72 http://www.deepdyve.com/lp/association-for-computing-machinery/long-short-term-memory-recurrent-neural-network-based-multimodal-PD8TcEBEm5
[70]	Sariyanidi E, Gunes H, Cavallaro A. Automatic analysis of facial affect:a survey of registration, representation, and recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(6):1113-1133 doi: 10.1109/TPAMI.2014.2366127
[71]	Lecun Y, Bengio Y, Hinton G. Deep learning. Nature, 2015, 521(7553):436-444 doi: 10.1038/nature14539
[72]	尹宝才, 王文通, 王立春.深度学习研究综述.北京工业大学学报, 2015, 41(1):48-59 http://d.old.wanfangdata.com.cn/Periodical/jsjyyyj201208002 Yin Bao-Cai, Wang Wen-Tong, Wang Li-Chun. Review of deep learning. Journal of Beijing University of Technology, 2015, 41(1):48-59 http://d.old.wanfangdata.com.cn/Periodical/jsjyyyj201208002
[73]	Zheng W Q, Yu J S, Zou Y X. An experimental study of speech emotion recognition based on deep convolutional neural networks. In: Proceedings of the 2015 International Conference on Affective Computing and Intelligent Interaction. Xi'an, China: IEEE, 2015. 827-831 http://ieeexplore.ieee.org/document/7344669/
[74]	Poria S, Chaturvedi I, Cambria E, Hussain A. Convolutional MKL based multimodal emotion recognition and sentiment analysis. In: Proceedings of the 16th IEEE International Conference on Data Mining. Barcelona, Spain: IEEE, 2016. 439-448 http://ieeexplore.ieee.org/document/7837868
[75]	Weninger F, Ringeval F, Marchi E, Schuller B. Discriminatively trained recurrent neural networks for continuous dimensional emotion recognition from audio. In: Proceedings of the 25th International Joint Conference on Artificial Intelligence. New York, USA: AAAI Press, 2016. 2196-2202 http://www.sewaproject.eu/files/338a1be0-486e-43ae-c09f-4e88236e62df.pdf
[76]	Banda N, Engelbrecht A, Robinson P. Continuous emotion recognition using a particle swarm optimized NARX neural network. In: Proceedings of the 2015 International Conference on Affective Computing and Intelligent Interaction. Xi'an, China: IEEE, 2015. 380-386 http://ieeexplore.ieee.org/document/7344599/
[77]	Glodek M, Tschechne S, Layher G, Schels M, Brosch T, Scherer S, et al. Multiple classifier systems for the classification of audio-visual emotional states. In: Proceedings of the 2011 International Conference on Affective Computing and Intelligent Interaction. Berlin, Heidelberg, German: Springer-Verlag, 2011. 359-368 http://www.springerlink.com/content/77345u71p6x76pg3/
[78]	Schuller B, Valstar M, Cowie R, Pantic M. AVEC 2012: the continuous audio/visual emotion challenge-an introduction. In: Proceedings of the 14th ACM International Conference on Multimodal Interaction. Santa Monica, California, USA: ACM, 2012. 361-362 http://www.mendeley.com/research/avec-2012-continuous-audiovisual-emotion-challenge/
[79]	Valstar M, Schuller B, Smith K, Almaev T, Eyben F, Krajewski J, et al. AVEC 2014: 3D dimensional affect and depression recognition challenge. In: Proceedings of the 4th International Workshop on Audio/Visual Emotion Challenge. Orlando, Florida, USA: ACM, 2014. 3-10 http://dl.acm.org/citation.cfm?id=2661807
[80]	Wu C H, Lin J C, Wei W L. Two-level hierarchical alignment for semi-coupled HMM-based audiovisual emotion recognition with temporal course. IEEE Transactions on Multimedia, 2013, 15(8):1880-1895 doi: 10.1109/TMM.2013.2269314
[81]	Mariooryad S, Busso C. Correcting time-continuous emotional labels by modeling the reaction lag of evaluators. IEEE Transactions on Affective Computing, 2015, 6(2):97-108 doi: 10.1109/TAFFC.2014.2334294
[82]	Mariooryad S, Busso C. Analysis and compensation of the reaction lag of evaluators in continuous emotional annotations. In: Proceedings of the 2013 Humaine Association Conference on Affective Computing and Intelligent Interaction. Geneva, Switzerland: IEEE, 2013. 85-90 http://dl.acm.org/citation.cfm?id=2544966
[83]	Schuller B, Valstar M, Eyben F, McKeown G, Cowie R, Pantic M. AVEC 2011-the first international audio/visual emotion challenge. In: Proceedings of the 2011 International Conference on Affective Computing and Intelligent Interaction. Berlin, German: Springer-Verlag, 2011. 415-424 AVEC 2011-the first international audio/visual emotion challenge.