Research map / Reinforcement learning / Visual and representation RL
Strategic Attentive Writer for Learning Macro-Actions
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 |
|---|---|---|---|
| add structured exploration (noisy Gaussian channel) to STRAW (ablation) | STRAW · Atari Frostbite | score: 4394 -> 8108 | Helped |
| add structured exploration to STRAW (ablation) | STRAW · Atari Q-bert | score: 20933 -> 23892 | Helped |
| add structured exploration to STRAW (ablation) | STRAW · Atari Amidar | score: 2223 -> 2022 | Hurt |
| add structured exploration to STRAW (ablation) | STRAW · Atari Alien | score: 2626 -> 3230 | Helped |
| replace LSTM with STRAWe | LSTM · Atari Frostbite | score: 1281 -> 8108 | Helped |
| replace LSTM with STRAWe | LSTM · Atari Ms. Pacman | score: 4227 -> 6902 | Helped |
| replace LSTM with STRAW | LSTM · Atari Breakout | score: 644 -> 344 | Hurt |
| replace FF baseline with STRAWe | FF · Atari Hero | score: 39929 -> 36931 | Hurt |
| replace FF with STRAWe | FF · Atari Ms. Pacman | score: 2014 -> 6902 | Helped |
| learn when to re-plan versus re-plan at every step or random commitment of 0 to 4 steps (ablation) | STRAW · Ms. Pacman ablation (100 epochs) | episode reward | Helped |
| increase Gaussian channel dimensionality (ablation) | STRAWe · Ms. Pacman ablation (100 epochs) | episode reward | Mixed |
| curriculum maze with increasing goal distance: STRAWe versus LSTM | STRAWe · 15x15 random mazes | episode reward | Helped |
Closest papers by meaning
- PcLast: Discovering Plannable Continuous Latent States (ICML 2024 · similarity 0.84)
- Embodied Task Planning via Graph-Informed Action Generation with Large Language Models (ICML 2026 · similarity 0.84)
- Learning Planning Abstractions from Language (ICLR 2024 · similarity 0.84)
- Procedural Memory Distillation: Online Reflection for Self-Improving Language Models (NeurIPS 2026 · similarity 0.83)
- Recurrent Reinforcement Learning with Memoroids (NeurIPS 2024 · similarity 0.83)
- OptionZero: Planning with Learned Options (ICLR 2025 · similarity 0.83)
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.