Research map / Reinforcement learning / Offline RL
Model-based Offline RL via Robust Value-Aware Model Learning with Implicitly Differentiable Adaptive Weighting
Zhongjian Qiao, Jiafei Lyu, Boxiang Lyu, Yao Shu, Siyang Gao, Shuang Qiu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace RAMBO with ROMI (robust value-aware model learning plus adaptive weighting) | RAMBO · D4RL MuJoCo (12 datasets, total) | total normalized score: 804.1 -> 953.5 | Improved |
| replace RAMBO with ROMI | RAMBO · NeoRL (9 datasets, total) | total normalized score: 382.8 -> 472.2 | Improved |
| replace MOBILE with ROMI | MOBILE · D4RL MuJoCo (12 datasets, total) | total normalized score: 857.7 -> 953.5 | Improved |
| replace Count-MORL with ROMI | Count-MORL · D4RL MuJoCo (12 datasets, total) | total normalized score: 927.5 -> 953.5 | Improved |
| replace RAMBO with ROMI on hopper-medium-replay | RAMBO · D4RL hopper-medium-replay-v2 | normalized score: 77.2 -> 102.0 | Improved |
| replace RAMBO with ROMI on walker2d-medium-expert | RAMBO · D4RL walker2d-medium-expert-v2 | normalized score: 73.7 -> 113.3 | Improved |
| replace RAMBO with ROMI on halfcheetah-medium | RAMBO · D4RL halfcheetah-medium-v2 | normalized score: 76.2 -> 76.2 | No clear effect |
| replace CQL with ROMI on NeoRL | CQL · NeoRL (9 datasets, total) | total normalized score: 466.3 -> 472.2 | Improved |
| add implicitly differentiable adaptive weighting to robust value-aware model learning (ablation) | ROMI · halfcheetah-medium-replay and hopper-medium-replay | model prediction error and normalized score | Improved |
| increase RAMBO adversarial coefficient lambda from 3e-4 to 5e-2 or 1e-1 (ablation) | RAMBO · halfcheetah-medium-expert-v2 | Q-value estimate: 3e-4 -> 5e-2, 1e-1 | Hurt |
| vary RAMBO lambda among 0, 3e-4, 5e-3 (ablation) | RAMBO · halfcheetah-medium-expert-v2 | Q-value estimate: 0, 3e-4, 5e-3 | No clear effect |
| increase uncertainty set scale xi from 0.01 to 10 (ablation) | ROMI · halfcheetah-medium-expert-v2 | Q-value estimate and training stability: 0.01, 0.1, 1.0, 10 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MOPO — MOPO: Model-based Offline Policy Optimization (NeurIPS 2020) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.91)
- Enhancing Value Function Estimation through First-Order State-Action Dynamics in Offline Reinforcement Learning (ICML 2024 · similarity 0.90)
- Model-based Offline Reinforcement Learning with Lower Expectile Q-Learning (ICLR 2025 · similarity 0.89)
- OCEAN-MBRL: Offline Conservative Exploration for Model-Based Offline Reinforcement Learning (AAAI 2024 · similarity 0.89)
- A Perspective of Q-value Estimation on Offline-to-Online Reinforcement Learning (AAAI 2024 · similarity 0.89)
- Adversarial Diffusion for Robust Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
