Research map / Reinforcement learning / Visual and representation RL
Learn What Not to Learn: Action Elimination with Deep Reinforcement Learning
Tom Zahavy, Matan Haroush, Nadav Merlis, Daniel J Mankowitz, Shie Mannor
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 |
|---|---|---|---|
| Use AE-DQN instead of prior work | Open Zork with A3 · Open Zork, A3 | cumulative reward: 13.5 -> 39 | Improved |
| Use AE-DQN instead of prior work | Open Zork with A4 · Open Zork, A4 | cumulative reward: 8.8 -> 16 | Improved |
| Train AE-DQN for 2M rather than 1M steps | Open Zork with A3 · Open Zork, A3 | cumulative reward: 39 -> 44 | Improved |
| Train AE-DQN for 2M rather than 1M steps | Open Zork with A4 · Open Zork, A4 | cumulative reward: 16 -> 16 | No clear effect |
| Add action elimination to Q-learning | Tabular Q-learning · Grid World | episode performance | Improved |
| Add action elimination to DQN | DQN · Troll Quest | cumulative reward | Improved |
| Use action elimination with the larger action space | AE-DQN versus DQN · Egg Quest, A2 with T=100 | learning speed | Improved |
| Increase the number of state categories, increasing invalid actions | Tabular action elimination Q-learning · Grid World | learning performance | Improved |
| Use an almost completely random elimination signal | Tabular action elimination Q-learning · Grid World | learning performance | No clear effect |
Closest papers by meaning
- No Prior Mask: Eliminate Redundant Action for Deep Reinforcement Learning (AAAI 2024 · similarity 0.87)
- Coarse-to-fine Q-Network with Action Sequence for Data-Efficient Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Adaptive $Q$-Network: On-the-fly Target Selection for Deep Reinforcement Learning (ICLR 2025 · similarity 0.85)
- Confounding Robust Deep Reinforcement Learning: A Causal Approach (NeurIPS 2025 · similarity 0.85)
- Deep Double Q-learning (NeurIPS 2026 · similarity 0.85)
- LipsNet++: Unifying Filter and Controller into a Policy Network (ICML 2025 · similarity 0.85)
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.
