Research map / Reinforcement learning / Offline RL
Peng's Q($\lambda$) for Conservative Value Estimation in Offline Reinforcement Learning
Byeongchan Kim, Min-hwan Oh
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 Bellman operator with Peng Q(lambda) in conservative value estimation | CQL · MuJoCo total (D4RL) | normalized score: 1010.2 -> 1252.1 | Improved |
| Replace Bellman operator with Peng Q(lambda) | CQL · AntMaze total (D4RL) | normalized score: 303.6 -> 397.6 | Improved |
| Replace Bellman operator with Peng Q(lambda) | CQL · Adroit total (D4RL) | normalized score: 93.6 -> 166.7 | Improved |
| CPQL vs strongest baselines on Adroit | CPQL · Adroit total | normalized score: 166.7 vs CSVE 168.5 | No clear effect |
| CPQL on antmaze-umaze-diverse | CPQL · antmaze-umaze-diverse | normalized score: 89.1 (MISA) vs 68.6 | Hurt |
| Use CPQL with small conservatism alpha (ablation) | CPQL vs CQL · D4RL datasets | normalized score | Improved |
| Use Peng Q(lambda) instead of uncorrected n-step, Retrace, Tree-backup (ablation) | CPQL · D4RL training curves | normalized score | Mixed |
| Pretrain Q with CPQL before online PQL fine-tuning | PQL · offline-to-online D4RL | return | Improved |
| Use multi-step operator on low-quality datasets | CPQL · low-quality datasets | performance | Hurt |
Closest papers by meaning
- A Perspective of Q-value Estimation on Offline-to-Online Reinforcement Learning (AAAI 2024 · similarity 0.93)
- Exclusively Penalized Q-learning for Offline Reinforcement Learning (NeurIPS 2024 · similarity 0.91)
- On the Complexity of Offline Reinforcement Learning with Q*-Approximation and Partial Coverage (NeurIPS 2026 · similarity 0.91)
- Model-based Offline Reinforcement Learning with Lower Expectile Q-Learning (ICLR 2025 · similarity 0.91)
- Enhancing Value Function Estimation through First-Order State-Action Dynamics in Offline Reinforcement Learning (ICML 2024 · similarity 0.90)
- Q-Distribution guided Q-learning for offline reinforcement learning: Uncertainty penalized Q-value via consistency model (NeurIPS 2024 · similarity 0.90)
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.
