Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Dynamic Action Repetition for Deep Reinforcement Learning
Aravind Lakshminarayanan, Sahil Sharma, Balaraman Ravindran
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 |
|---|---|---|---|
| add dynamic action repetition (ARR 4 and 20) to DQN | DQN · Seaquest | average score: 5450 -> 10458 | Improved |
| add dynamic action repetition to DQN | DQN · Space Invaders | average score: 1189 -> 2796 | Improved |
| add dynamic action repetition to DQN | DQN · Alien | average score: 2085 -> 3114 | Improved |
| add dynamic action repetition to DQN | DQN · Enduro | average score: 707 -> 1002 | Improved |
| add dynamic action repetition to A3C | A3C · Space Invaders | average score: 603.04 -> 2220.18 | Improved |
| add dynamic action repetition to A3C | A3C · Enduro | average score: 0.77 -> 497.29 | Improved |
| add dynamic action repetition to A3C | A3C · Alien | average score: 1440.82 -> 2722.19 | Improved |
| add dynamic action repetition to A3C | A3C · Seaquest | average score: 1769.12 -> 2047.74 | Improved |
| add dynamic action repetition to A3C | A3C · Q*bert | average score: 21184.75 -> 21405.35 | Improved |
| enlarge DQN hidden layer from 512 to 1024 units (control for capacity) (ablation) | DQN · Seaquest | average score: 5860 -> 5450 | No clear effect |
| use higher static ARR 20 instead of dynamic ARR (ablation) | DFDQN · Seaquest | average score: 10458 -> 1707 | Hurt |
| use higher static ARR 20 instead of dynamic ARR (ablation) | DFDQN · Enduro | average score: 1002 -> 124 | Hurt |
Closest papers by meaning
- Improving planning and MBRL with temporally-extended actions (NeurIPS 2025 · similarity 0.86)
- Overcoming Slow Decision Frequencies in Continuous Control: Model-Based Sequence Reinforcement Learning for Model-Free Control (ICLR 2025 · similarity 0.86)
- Select before Act: Spatially Decoupled Action Repetition for Continuous Control (ICLR 2025 · similarity 0.86)
- Deep Reinforcement Learning with Time-Scale Invariant Memory (AAAI 2025 · similarity 0.85)
- No Prior Mask: Eliminate Redundant Action for Deep Reinforcement Learning (AAAI 2024 · similarity 0.85)
- Enabling Realtime Reinforcement Learning at Scale with Staggered Asynchronous Inference (ICLR 2025 · similarity 0.84)
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.
