paper-with-me

홈 › Papers

When to Update Your Model: Constrained Model-based Reinforcement Learning

2022-10-15 · Tianying Ji, Yu Luo, Fuchun Sun, Mingxuan Jing, Fengxiang He, Wenbing Huang

Designing and analyzing model-based RL (MBRL) algorithms with guaranteed monotonic improvement has been challenging, mainly due to the interdependence between policy optimization and model learning. Existing discrepancy bounds generally ignore the impacts of model shifts, and their corresponding algorithms are prone to degrade performance by drastic model updating. In this work, we first propose a novel and general theoretical scheme for a non-decreasing performance guarantee of MBRL. Our follow-up derived bounds reveal the relationship between model shifts and performance improvement. These discoveries encourage us to formulate a constrained lower-bound optimization problem to permit the monotonicity of MBRL. A further example demonstrates that learning models from a dynamically-varying number of explorations benefit the eventual returns. Motivated by these analyses, we design a simple but effective algorithm CMLO (Constrained Model-shift Lower-bound Optimization), by introducing an event-triggered mechanism that flexibly determines when to update the model. Experiments show that CMLO surpasses other state-of-the-art methods and produces a boost when various policy optimization methods are employed.

📄 PDF Abstract BibTeX arXiv:2210.08349

Code (1)

jity16/when-to-update-your-model-constrained-model-based-reinforcement-learning 공식 구현 pytorch

Tasks

modelModel-based Reinforcement Learningreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Similar Papers 제목 키워드 기반

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

2025-10-25 · Guojian Zhan, Likun Wang, Pengcheng Wang, Feihong Zhang 외 arxiv

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-stationar…

Reinforcement LearningOpenAI Gym

Selection as Power: Constrained Reinforcement for Bounded Decision Authority

2026-03-02 · Jose Manuel de la Chica Rodriguez, Juan Manuel Vera Díaz arxiv

Selection as Power argued that upstream selection authority, rather than internal objective misalignment, constitutes a primary source of risk in high-stakes agentic systems. However, the original framework was static: g…

Anytime Safe Reinforcement Learning

2025-04-23 · Pol Mestres, Arnau Marzabal, Jorge Cortés

This paper considers the problem of solving constrained reinforcement learning problems with anytime guarantees, meaning that the algorithmic solution returns a safe policy regardless of when it is terminated. Drawing in…

reinforcement-learningReinforcement LearningSafe Reinforcement Learning

Regression Bugs Are In Your Model! Measuring, Reducing and Analyzing Regressions In NLP Model Updates

2021-05-07 · ACL 2021 5 · Yuqing Xie, Yi-An Lai, Yuanjun Xiong, Yi Zhang 외

Behavior of deep neural networks can be inconsistent between different versions. Regressions during model update are a common cause of concern that often over-weigh the benefits in accuracy or efficiency gain. This work …

Knowledge Distillationmodelnegative flip rateregression

SB-TRPO: Towards Safe Reinforcement Learning with Hard Constraints

2025-12-29 · Dominik Wagner, Ankit Kanwar, Luke Ong arxiv

In safety-critical domains, reinforcement learning (RL) agents must often satisfy strict, zero-cost safety constraints while accomplishing tasks. Existing model-free methods frequently either fail to achieve near-zero sa…

Reinforcement Learning