首页分享触觉手势情感识别的超限学习方法

触觉手势情感识别的超限学习方法

来源：萌宠菠菠乐园时间：2024-10-18 09:30

[1]魏佳琪,刘华平,王博文,等.触觉手势情感识别的超限学习方法[J].智能系统学报,2019,14(1):127-133.[doi:10.11992/tis.201804029]
　WEI Jiaqi,LIU Huaping,WANG Bowen,et al.Extreme learning machine for emotion recognition of tactile gestures[J].CAAI Transactions on Intelligent Systems,2019,14(1):127-133.[doi:10.11992/tis.201804029]

点击复制

《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷: 14 期数: 2019年第1期页码: 127-133 栏目: 学术论文—机器感知与模式识别出版日期: 2019-01-05

Title: Extreme learning machine for emotion recognition of tactile gestures 作者: 1. 河北工业大学省部共建电工装备可靠性与智能化国家重点实验室, 天津 300130;
2. 清华大学智能技术与系统国家重点实验室, 北京 100084 Author(s): WEI Jiaqi1, LIU Huaping2, WANG Bowen1, SUN Fuchun2 1. State Key Laboratory of Reliability and Intelligence of Electrical Equipment, Hebei University of Technology, Tianjin 300130, China;
2. State Key Lab. of Intelligent Technology and Systems, Tsinghua University, Beijing 100084, China 关键词: 触觉; 情感识别; 极限学习机; 特征提取; 触摸手势; 支持向量机; 人机交互; 机器学习 Keywords: haptic; emotion recognition; extreme learning machine; feature extraction; touch gesture; support vector machine; human-computer interaction; machine learning 分类号: TP391.4 DOI: 10.11992/tis.201804029 摘要: 为了解决声音和图像情感识别的不足，提出一种新的情感识别方式：触觉情感识别。对CoST（corpus of social touch）数据集进行了一系列触觉情感识别研究，对CoST数据集进行数据预处理，提出一些关于触觉情感识别的特征。利用极限学习机分类器探究不同手势下的情感识别，对14种手势下的3种情感（温柔、正常、暴躁）进行识别，准确度较高，且识别速度快识别时间短。结果表明，手势的不同会影响情感识别的准确率，其中手势“stroke”的识别效果在不同分类器下的分类精度均为最高，且有较好的分类精度，达到72.07%；极限学习机作为触觉情感识别的分类器，具有较好的分类效果，识别速度快；有的手势本身对应着某种情感，从而影响分类结果。 Abstract: To overcome the deficiencies of sound and image emotion recognition, a new emotion recognition method, haptic emotion recognition, is proposed. A series of haptic emotion recognition studies on Corpus of Social Touch (CoST) datasets were performed. First, the CoST data was preprocessed, presenting some features about haptic emotion recognition. Using the extreme learning machine classifier to explore emotion recognition under different gestures, three kinds of emotions, gentle, normal, and irritable, under 14 kinds of gestures, were identified with higher accuracy and a faster recognition speed (0.04 s). The results showed that differences in gestures will affect the accuracy of emotion recognition, wherein the recognition effect of the gesture "stroke" is the highest in classification accuracy under different classifiers. This new method yielded better classification accuracy, reaching 72.07%. As a classifier of haptic emotion recognition, the extreme learning machine had better classification effect and faster recognition speed. Some gestures corresponded to certain emotions, which affected the classification results.

备注/Memo

收稿日期:2018-04-18。
基金项目:国家自然科学基金重点项目（U1613212）；河北省自然科学基金项目（E2017202035）.
作者简介:魏佳琪,男,1995年生,硕士研究生,主要研究方向为新型磁性材料与器件、触觉交互;刘华平,男,1976年生,副教授,博士生导师,IEEE Senior Member、中国人工智能学会理事,中国人工智能学会认知系统与信息处理专业委员会秘书长,主要研究方向为机器人感知、学习与控制、多模态信息融合。在IEEE Trans.On Automatic Control、IEEE Trans.on Circuits and Systems Ⅱ以及Automatica等国际期刊,以及ICRA、IROS等国际会议中发表论文十余篇;王博文,男,1956年生,教授,博士生导师,主要研究方向为磁致伸缩材料与器件、振动发电技术、磁特性测试技术。发表学术论文200余篇,被SCI收录100余篇。
通讯作者:刘华平.E-mail:hpliu@tsinghua.edu.cn