Research map / Reinforcement learning / MDPs and planning
A Max-Min Entropy Framework for Reinforcement Learning
Seungyul Han, Youngchul Sung
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 the future-entropy value objective with a reversed-entropy objective | SAC in the 4-room maze · 4-room maze | number of visited states | Improved |
| Use the max-min entropy objective instead of SAC's maximum-entropy objective | SAC · Sparse Hopper | max average return: 823.70$\pm$215.35 -> 902.50$\pm$4.36 | Improved |
| Use the max-min entropy objective instead of SAC's maximum-entropy objective | SAC · Delayed HalfCheetah | max average return: 3742.33$\pm$3064.55 -> 8451.20$\pm$1375.27 | Improved |
| Use the max-min entropy objective instead of SAC's maximum-entropy objective | SAC · HumanoidStandup | max average return: 167394.36 $\pm$7291.99 -> 267734.03 $\pm$74302.99 | Improved |
| Add disentangled exploration and exploitation (ablation) | MME · Delayed HalfCheetah | max average return: 7299.28$\pm$1562.19 -> 8451.20$\pm$1375.27 | Improved |
| Increase αQ from too small a setting toward the selected αQ=2.0 (ablation) | DE-MME · Delayed HalfCheetah | max average return | Mixed |
| Increase απ from too small a setting toward the selected απ=0.2 (ablation) | DE-MME · Delayed HalfCheetah | max average return | Mixed |
| Use the minibatch-based positive offset for negative entropy instead of the maximum-entropy constant offset (ablation) | MME · pure exploration | exploration | Improved |
Closest papers by meaning
- Maximum Entropy Reinforcement Learning with Diffusion Policy (ICML 2025 · similarity 0.90)
- Maximum Entropy Reinforcement Learning via Energy-Based Normalizing Flow (NeurIPS 2024 · similarity 0.90)
- S$2$AC: Energy-Based Reinforcement Learning with Stein Soft Actor Critic (ICLR 2024 · similarity 0.90)
- When Maximum Entropy Misleads Policy Optimization (ICML 2025 · similarity 0.89)
- DIME: Diffusion-Based Maximum Entropy Reinforcement Learning (ICML 2025 · similarity 0.89)
- Maximum Entropy Heterogeneous-Agent Reinforcement Learning (ICLR 2024 · similarity 0.88)
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.
