paper-with-me

홈 › Papers

Revisiting Design Choices in Offline Model Based Reinforcement Learning

2021-05-21 · NeurIPS 2021 12 · Cong Lu, Philip Ball, Jack Parker-Holder, Michael Osborne, S Roberts

Offline reinforcement learning enables agents to make use of large pre-collected datasets of environment transitions and learn control policies without the need for potentially expensive or unsafe online data collection. Recently, significant progress has been made in offline RL, with the dominant approach becoming methods which leverage a learned dynamics model. This typically involves constructing a probabilistic model, and using it to penalize rewards in regions of high uncertainty, solving for a pessimistic MDP that lower bounds the true MDP. Recent work, however, exhibits a breakdown between theory and practice, whereby pessimistic return ought to be bounded by the total variation distance of the model from the true dynamics, but is instead implemented through a penalty based on estimated model uncertainty. This has spawned a variety of uncertainty heuristics, with little to no comparison between differing approaches. In this paper, we show these heuristics have significant interactions with other design choices, such as the number of models in the ensemble, the model rollout length and the penalty weight. Furthermore, we compare these uncertainty heuristics under a new evaluation protocol that, for the first time, captures the specific covariate shift induced by model-based RL. This allows us to accurately assess the calibration of different proposed penalties. Finally, with these insights, we show that selecting these key hyperparameters using Bayesian Optimization produces drastically stronger performance than existing hand-tuned methods.

📄 PDF Abstract BibTeX

Code (0)

등록된 구현이 없습니다.

Tasks

Bayesian OptimizationModel-based Reinforcement LearningOffline RLreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Similar Papers 제목 키워드 기반

Revisiting the Minimalist Approach to Offline Reinforcement Learning

2023-05-16 · NeurIPS 2023 11 · Denis Tarasov, Vladislav Kurenkov, Alexander Nikulin, Sergey Kolesnikov

Recent years have witnessed significant advancements in offline reinforcement learning (RL), resulting in the development of numerous algorithms with varying degrees of complexity. While these algorithms have led to note…

D4RLOffline RLreinforcement-learningReinforcement Learning+1

Revisiting Design Choices in Offline Model-Based Reinforcement Learning

2021-10-08 · Cong Lu, Philip J. Ball, Jack Parker-Holder, Michael A. Osborne 외

Offline reinforcement learning enables agents to leverage large pre-collected datasets of environment transitions to learn control policies, circumventing the need for potentially expensive or unsafe online data collecti…

Bayesian OptimizationModel-based Reinforcement Learningreinforcement-learningReinforcement Learning+1

Revisiting Design Choices in Proximal Policy Optimization

2020-09-23 · Chloe Ching-Yun Hsu, Celestine Mendler-Dünner, Moritz Hardt

Proximal Policy Optimization (PPO) is a popular deep policy gradient algorithm. In standard implementations, PPO regularizes policy updates with clipped probability ratios, and parameterizes policies with either continuo…

MuJoCo

The Three Regimes of Offline-to-Online Reinforcement Learning

2025-10-01 · Lu Li, Tianwei Ni, Yihao Sun, Pierre-Luc Bacon arxiv

Offline-to-online reinforcement learning (RL) has emerged as a practical paradigm that leverages offline datasets for pretraining and online interactions for fine-tuning. However, its empirical behavior is highly inconsi…

Reinforcement Learning

Adversarial Model for Offline Reinforcement Learning

2023-02-21 · NeurIPS 2023 11 · Mohak Bhardwaj, Tengyang Xie, Byron Boots, Nan Jiang 외

We propose a novel model-based offline Reinforcement Learning (RL) framework, called Adversarial Model for Offline Reinforcement Learning (ARMOR), which can robustly learn policies to improve upon an arbitrary reference …

modelreinforcement-learningReinforcement LearningReinforcement Learning (RL)