Research map / Reinforcement learning / Offline RL
EMaQ: Expected-Max Q-Learning Operator for Simple Yet Effective Offline and Online RL
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use EMaQ instead of the strongest listed baseline | Offline RL · kitchen-complete, kitchen-partial, kitchen-mixed, antmaze-umaze, antmaze-umaze-diverse | normalized score: 27.2 ± 3.2 -> 36.9 ± 3.7; 69.3 ± 5.2 -> 74.6 ± 0.6; 65.5 ± 1.8 -> 70.8 ± 2.3; 59.0 ± 5.5 -> 91.0 ± 4.6; 68.0 ± 19.0 -> 9 | Helped |
| Use EMaQ instead of the strongest listed baseline | Offline RL · antmaze-medium-play, antmaze-medium-diverse | normalized score: 3.5 ± 6.1 -> 0.0 ± 0.0; 0.4 ± 0.8 -> 0.0 ± 0.0 | Hurt |
| Use EMaQ instead of the strongest listed baseline | Offline RL · antmaze-large-play, antmaze-large-diverse | normalized score: 0.0 ± 0.0 -> 0.0 ± 0.0; 0.0 ± 0.0 -> 0.0 ± 0.0 | No clear effect |
| Use EMaQ instead of the strongest listed baseline | Offline RL · door-cloned, hammer-cloned, pen-cloned, relocate-cloned | normalized score: 0.2 ± 0.4 -> 0.2 ± 0.3; 1.3 ± 0.5 -> 1.0 ± 0.7; 43.8 ± 6.4 -> 27.9 ± 3.7; 0.0 ± 0.0 -> -0.2 ± 0.2 | Mixed |
| Increase the number of sampled actions N (ablation) | EMaQ · D4RL Mujoco offline tasks | return | Helped |
| Increase N in medium-expert settings (ablation) | EMaQ · D4RL medium-expert tasks | return | Hurt |
| Replace the autoregressive behavior model with a VAE (ablation) | EMaQ · D4RL offline tasks | return | Hurt |
| Add a distilled proposal distribution and sample from a mixture with the behavior model (ablation) | EMaQ · Mujoco offline tasks | return | No clear effect |
| Use an autoregressive proposal model rather than the prior VAE design (ablation) | Offline EMaQ · D4RL tasks | return | Helped |
| Remove the BCQ action-perturbation network and select among sampled behavior-policy actions (ablation) | BCQ-derived offline RL · D4RL offline tasks | normalized score | Mixed |
Closest papers by meaning
- On the Complexity of Offline Reinforcement Learning with Q*-Approximation and Partial Coverage (NeurIPS 2026 · similarity 0.90)
- Model-based Offline Reinforcement Learning with Lower Expectile Q-Learning (ICLR 2025 · similarity 0.90)
- Offline-Boosted Actor-Critic: Adaptively Blending Optimal Historical Behaviors in Deep Off-Policy RL (ICML 2024 · similarity 0.90)
- A Perspective of Q-value Estimation on Offline-to-Online Reinforcement Learning (AAAI 2024 · similarity 0.90)
- A Clean Slate for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Exclusively Penalized Q-learning for Offline Reinforcement Learning (NeurIPS 2024 · 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.