Play with Emotion: Affect-Driven Reinforcement Learning

被引:4
作者
Barthet, Matthew [1 ]
Khalifa, Ahmed [1 ]
Liapis, Antonios [1 ]
Yannakakis, Georgios N. [1 ]
机构
[1] Univ Malta, Inst Digital Games, Msida, Malta
来源
2022 10TH INTERNATIONAL CONFERENCE ON AFFECTIVE COMPUTING AND INTELLIGENT INTERACTION (ACII) | 2022年
关键词
Reinforcement Learning; Arousal; Go-Blend; Go-Explore; Affective Computing; Gameplaying; MODELS;
D O I
10.1109/ACII55700.2022.9953887
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
This paper introduces a paradigm shift by viewing the task of affect modeling as a reinforcement learning (RL) process. According to the proposed paradigm, RL agents learn a policy (i.e. affective interaction) by attempting to maximize a set of rewards (i.e. behavioral and affective patterns) via their experience with their environment (i.e. context). Our hypothesis is that RL is an effective paradigm for interweaving affect elicitation and manifestation with behavioral and affective demonstrations. Importantly, our second hypothesis-building on Damasio's somatic marker hypothesis-is that emotion can be the facilitator of decision-making. We test our hypotheses in a racing game by training Go-Blend agents to model human demonstrations of arousal and behavior; Go-Blend is a modified version of the Go-Explore algorithm which has recently showcased supreme performance in hard exploration tasks. We first vary the arousal-based reward function and observe agents that can effectively display a palette of affect and behavioral patterns according to the specified reward. Then we use arousal-based state selection mechanisms in order to bias the strategies that Go-Blend explores. Our findings suggest that Go-Blend not only is an efficient affect modeling paradigm but, more importantly, affect-driven RL improves exploration and yields higher performing agents, validating Damasio's hypothesis in the domain of games.
引用
收藏
页数:8
相关论文
共 37 条
[1]  
Ammanabrolu P, 2020, Arxiv, DOI arXiv:2002.08795
[2]   Deceptive Games [J].
Anderson, Damien ;
Stephenson, Matthew ;
Togelius, Julian ;
Salge, Christoph ;
Levine, John ;
Renz, Jochen .
APPLICATIONS OF EVOLUTIONARY COMPUTATION, EVOAPPLICATIONS 2018, 2018, 10784 :376-391
[3]  
Barthet M., 2022, PROC FDN DIGITAL GAM
[4]  
Barthet M., 2021, PROC IEEE C AFFECTIV
[5]   The role of emotion in decision-making: Evidence from neurological patients with orbitofrontal damage [J].
Bechara, A .
BRAIN AND COGNITION, 2004, 55 (01) :30-40
[6]  
Berner Christopher, 2019, arXiv
[7]  
Broekens J, 2007, LECT NOTES COMPUT SC, V4527, P357, DOI 10.1007/978-3-540-73053-8_36
[8]   Affect Detection: An Interdisciplinary Review of Models, Methods, and Their Applications [J].
Calvo, Rafael A. ;
D'Mello, Sidney .
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING, 2010, 1 (01) :18-37
[9]   The somatic marker hypothesis and the possible functions of the prefrontal cortex [J].
Damasio, AR .
PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY B-BIOLOGICAL SCIENCES, 1996, 351 (1346) :1413-1420
[10]  
DUDANI SA, 1976, IEEE T SYST MAN CYB, V6, P327