Human-in-the-loop Reinforcement Learning for Emotion Recognition

被引:0
作者
Tan, Swee Yang [1 ]
Yau, Kok-Lim Alvin [1 ]
机构
[1] Univ Tunku Abdul Rahman UTAR, Lee Kong Chian Fac Engn & Sci, Kajang, Selangor, Malaysia
来源
2024 IEEE 14TH SYMPOSIUM ON COMPUTER APPLICATIONS & INDUSTRIAL ELECTRONICS, ISCAIE 2024 | 2024年
关键词
reinforcement learning; human-in-the-loop; augmented intelligence; emotion recognition;
D O I
10.1109/ISCAIE61308.2024.10576433
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Facial emotion recognition (FER) analyses a person's facial expressions in images to understand the person's emotions. Achieving high accuracy in real-world FER systems requires extensive data, which poses challenges in cost, practicability, and privacy, especially in ensuring fairness and avoiding discrimination related to skin color or health conditions. This paper introduces a novel approach called two-state Q-learning with human feedback (TS-QL-HF) to enhance the accuracy of a FER system by integrating human feedback collected during the evaluation process as a refined reward function into two-state Q-learning (TS-QL) and double Q-learning (DQL) to operate in deterministic environments. This approach does not need an extensive dataset to include minority populations. Simulation results demonstrate that TS-QL-HF provides the highest accuracy. However, the tuned TS-QL, without human feedback, shows a higher efficiency despite a slightly lower accuracy due to the inaccurate reward function.
引用
收藏
页码:21 / 26
页数:6
相关论文
共 17 条
[1]   The Human in Emotion Recognition on Social Media: Attitudes, Outcomes, Risks [J].
Andalibi, Nazanin ;
Buss, Justin .
PROCEEDINGS OF THE 2020 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI'20), 2020,
[2]   Active Object Localization with Deep Reinforcement Learning [J].
Caicedo, Juan C. ;
Lazebnik, Svetlana .
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, :2488-2496
[3]   Integrating Machine Learning with Human Knowledge [J].
Deng, Changyu ;
Ji, Xunbi ;
Rainey, Colton ;
Zhang, Jianyu ;
Lu, Wei .
ISCIENCE, 2020, 23 (11)
[4]  
Deng J, 2009, PROC CVPR IEEE, P248, DOI 10.1109/CVPRW.2009.5206848
[5]  
Hafiz A. M., 2022, Handbook of Intelligent Computing and Optimization for Sustainable Development, P271
[6]   Deep Residual Learning for Image Recognition [J].
He, Kaiming ;
Zhang, Xiangyu ;
Ren, Shaoqing ;
Sun, Jian .
2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2016, :770-778
[7]  
Ilona T, Deep Reinforcement Learning in emotion recognition
[8]   A Review on Interactive Reinforcement Learning From Human Social Feedback [J].
Lin, Jinying ;
Ma, Zhen ;
Gomez, Randy ;
Nakamura, Keisuke ;
He, Bo ;
Li, Guangliang .
IEEE ACCESS, 2020, 8 :120757-120765
[9]  
Maroti A, 2019, Arxiv, DOI arXiv:1910.13701
[10]  
Rhue L., 2018, Racial influence on automated perceptions of emotions, DOI [DOI 10.2139/SSRN.3281765, 10.2139/ssrn.3281765]