DSAC: Distributional Soft Actor-Critic for Risk-Sensitive Reinforcement Learning
Journal of Artificial Intelligence Research · 25.06.2025 · 83
DOI: 10.1613/jair.1.17526 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Artificial Intelligence Research, online erschienen am 25.06.2025. Verfasst von Xiaoteng Ma und 5 weiteren Autor:innen. Die Arbeit wurde bisher 27-mal zitiert (Crossref).
Abstract (Originalsprache)
We present Distributional Soft Actor-Critic (DSAC), a distributional reinforcement learning (RL) algorithm that combines the strengths of distributional information of accumulated rewards and entropy-driven exploration from Soft Actor-Critic (SAC) algorithm. DSAC models the randomness in both action and rewards, surpassing baseline performances on various continuous control tasks. Unlike standard approaches that solely maximize expected rewards, we propose a unified framework for risk-sensitive learning, one that optimizes the risk-related objective while balancing entropy to encourage exploration. Extensive experiments demonstrate DSAC’s effectiveness in enhancing agent performances for both risk-neutral and risk-sensitive control tasks.
Zitieren
APA
Ma, X., Chen, J., Xia, L., Yang, J., Zhao, Q., & Zhou, Z. (2025). DSAC: Distributional Soft Actor-Critic for Risk-Sensitive Reinforcement Learning. Journal of Artificial Intelligence Research, 83. https://doi.org/10.1613/jair.1.17526
Vancouver
Ma X, Chen J, Xia L, Yang J, Zhao Q, Zhou Z. DSAC: Distributional Soft Actor-Critic for Risk-Sensitive Reinforcement Learning. Journal of Artificial Intelligence Research. 2025;83. doi:10.1613/jair.1.17526
BibTeX
@article{Ma2025,
author = {Ma, Xiaoteng and Chen, Junyao and Xia, Li and Yang, Jun and Zhao, Qianchuan and Zhou, Zhengyuan},
title = {DSAC: Distributional Soft Actor-Critic for Risk-Sensitive Reinforcement Learning},
journal = {Journal of Artificial Intelligence Research},
year = {2025},
volume = {83},
doi = {10.1613/jair.1.17526}
}
EUVIMED