| 引用本文: | 韩晶,解仑,王志良,任福继.基于GMM的增量式情感映射[J].哈尔滨工业大学学报,2018,50(8):168.DOI:10.11918/j.issn.0367-6234.201707059 |
| HAN Jing,XIE Lun,WANG Zhiliang,REN Fuji.Incrementally emotion mapping based on GMM[J].Journal of Harbin Institute of Technology,2018,50(8):168.DOI:10.11918/j.issn.0367-6234.201707059 |
|
| 摘要: |
| 为有效地获得用户的真实情感状态，促进和谐的人机交互体验.结合AVS情感空间和大五人格理论，提出一种基于高斯混合模型的增量式情感映射模型.首先，在AVS情感空间的3种属性(A, V, S)坐标轴上，利用高斯混合模型对情感类型进行依次建模，计算情感概率值及其空间分布；其次，针对用户的个体差异性，采用层次分析法研究人格五因素与情感属性之间的关联，获得用户的个性化认知参数，实现具有个性化认知的情感映射结果；之后，采用增量式学习方法对情感类型的分布空间进行实时修正，保证情感分类的高准确率.最后，实验结果验证了该方法的情感映射结果与用户的真实情感状态具有高度一致性，并有较好的自适应性.
|
| 关键词: 情感映射 AVS情感空间 高斯混合模型 大五人格理论 增量式学习 |
| DOI：10.11918/j.issn.0367-6234.201707059 |
| 分类号:TP751.1 |
| 文献标识码:A |
| 基金项目:国家重点研发计划重点专项课题(2016YFB1001404); 国家自然科学基金面上资助项目(61672093); 国家自然科学基金重点资助项目(61432004) |
|
| Incrementally emotion mapping based on GMM |
|
HAN Jing1,XIE Lun1,WANG Zhiliang1,REN Fuji2
|
|
(1. School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China; 2. Affective Computing and Advanced Intelligent Machines Key Laboratory(Hefei University of Technology), Hefei 230009, China)
|
| Abstract: |
| In order to obtain users' actual emotional status effectively and promote a harmonious human-computer interaction experience, combined with the AVS emotional space and big five personality theory, this paper proposes an incremental emotion mapping model based on Gauss mixture model. First of all, with three attributes in AVS emotional space (A, V, S) coordinate, the emotional probability value and space distribution is calculated with Gauss mixture model. Secondly, based on differences of individual users, analytic hierarchy method was used to study the relationship between big five personality and emotional attributes, personalized cognitive parameters of the user was obtained, and the realization of emotional mapping results with personalized knowledge were achieved. Then, incremental learning method was applied to get real-time correction of the spatial distribution of emotion type, thus ensuring the high accuracy of emotional classification. Finally, the experimental results show that this method has a high degree of consistency with the real emotional state of the user, and also has good adaptability.
|
| Key words: emotion mapping AVS emotional space Gaussian mixture model five-factor model incremental learning |
