Research map / Reinforcement learning / Visual and representation RL
Reuse of Neural Modules for General Video Game Playing
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 |
|---|---|---|---|
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Asterix | score: 2800 -> $3355_{bow}$ | Helped |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Bowling | score: 249.3 -> $265.0_{fr}$ | Helped |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Freeway | score: 31.4 -> $32.2_{brk}$ | Helped |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Boxing | score: 93.9 -> $95.0_{sea}$ | Helped |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Space Invaders | score: 1438.0 -> $1655.0_{po}$ | Helped |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Seaquest | score: 466.0 -> $975.0_{sp}$ | Helped |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Breakout | score: 31.0 -> $30.3_{box}$ | Hurt |
| Reuse a trained source network instead of training from scratch | GRUSM-ESP · Pong | score: 0.0 -> $10.0_{ast}$ | Mixed |
| Use GRUSM-ESP transfer rather than the random-network control | GRUSM-ESP · Pong | score: 21.0 -> $10.0_{ast}$ | Hurt |
| Use GRUSM-ESP transfer rather than the random-network control | GRUSM-ESP · Seaquest | score: 460.0 -> $975.0_{sp}$ | Helped |
| Use target complexity and training-complexity indicators to predict transfer effectiveness | GRUSM-ESP transfer · Atari source-target pairs vs. scratch | correlation R: R = 0.40; p-value < 0.0025 | Helped |
| Use target-complexity indicators to predict transfer effectiveness | GRUSM-ESP transfer · Atari source-target pairs vs. random | correlation R: R = 0.53; p-value < 10^{-7} | Helped |
Closest papers by meaning
- Multi-Game Decision Transformers (NeurIPS 2022 · similarity 0.87)
- Investigating Pre-Training Objectives for Generalization in Vision-Based Reinforcement Learning (ICML 2024 · similarity 0.86)
- End-to-End Neuro-Symbolic Reinforcement Learning with Textual Explanations (ICML 2024 · similarity 0.85)
- Towards Generalizable Reinforcement Learning via Causality-Guided Self-Adaptive Representations (ICLR 2025 · similarity 0.85)
- Efficient Reinforcement Learning by Discovering Neural Pathways (NeurIPS 2024 · similarity 0.85)
- BlendRL: A Framework for Merging Symbolic and Neural Policy Learning (ICLR 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.