Mind Your Entropy: From Maximum Entropy to Trajectory Entropy-Constrained RL

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhan, Guojian, Wang, Likun, Wang, Pengcheng, Zhang, Feihong, Duan, Jingliang, Tomizuka, Masayoshi, Li, Shengbo Eben
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908654196752384
author Zhan, Guojian
Wang, Likun
Wang, Pengcheng
Zhang, Feihong
Duan, Jingliang
Tomizuka, Masayoshi
Li, Shengbo Eben
author_facet Zhan, Guojian
Wang, Likun
Wang, Pengcheng
Zhang, Feihong
Duan, Jingliang
Tomizuka, Masayoshi
Li, Shengbo Eben
contents Maximum entropy has become a mainstream off-policy reinforcement learning (RL) framework for balancing exploitation and exploration. However, two bottlenecks still limit further performance improvement: (1) non-stationary Q-value estimation caused by jointly injecting entropy and updating its weighting parameter, i.e., temperature; and (2) short-sighted local entropy tuning that adjusts temperature only according to the current single-step entropy, without considering the effect of cumulative entropy over time. In this paper, we extends maximum entropy framework by proposing a trajectory entropy-constrained reinforcement learning (TECRL) framework to address these two challenges. Within this framework, we first separately learn two Q-functions, one associated with reward and the other with entropy, ensuring clean and stable value targets unaffected by temperature updates. Then, the dedicated entropy Q-function, explicitly quantifying the expected cumulative entropy, enables us to enforce a trajectory entropy constraint and consequently control the policy long-term stochasticity. Building on this TECRL framework, we develop a practical off-policy algorithm, DSAC-E, by extending the state-of-the-art distributional soft actor-critic with three refinements (DSAC-T). Empirical results on the OpenAI Gym benchmark demonstrate that our DSAC-E can achieve higher returns and better stability.
format Preprint
id arxiv_https___arxiv_org_abs_2511_11592
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mind Your Entropy: From Maximum Entropy to Trajectory Entropy-Constrained RL
Zhan, Guojian
Wang, Likun
Wang, Pengcheng
Zhang, Feihong
Duan, Jingliang
Tomizuka, Masayoshi
Li, Shengbo Eben
Machine Learning
Artificial Intelligence
Maximum entropy has become a mainstream off-policy reinforcement learning (RL) framework for balancing exploitation and exploration. However, two bottlenecks still limit further performance improvement: (1) non-stationary Q-value estimation caused by jointly injecting entropy and updating its weighting parameter, i.e., temperature; and (2) short-sighted local entropy tuning that adjusts temperature only according to the current single-step entropy, without considering the effect of cumulative entropy over time. In this paper, we extends maximum entropy framework by proposing a trajectory entropy-constrained reinforcement learning (TECRL) framework to address these two challenges. Within this framework, we first separately learn two Q-functions, one associated with reward and the other with entropy, ensuring clean and stable value targets unaffected by temperature updates. Then, the dedicated entropy Q-function, explicitly quantifying the expected cumulative entropy, enables us to enforce a trajectory entropy constraint and consequently control the policy long-term stochasticity. Building on this TECRL framework, we develop a practical off-policy algorithm, DSAC-E, by extending the state-of-the-art distributional soft actor-critic with three refinements (DSAC-T). Empirical results on the OpenAI Gym benchmark demonstrate that our DSAC-E can achieve higher returns and better stability.
title Mind Your Entropy: From Maximum Entropy to Trajectory Entropy-Constrained RL
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.11592