Research map / Detection and segmentation / Open-set and few-shot detection
Structured World Belief for Reinforcement Learning in POMDP
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 |
|---|---|---|---|
| use structured object-centric belief instead of unstructured AESMC belief | world model for supervised reasoning · 2D Maze supervised reasoning, K=10 | accuracy: 54.2 ± 5.2 -> 63.4 ± 2.6 | Helped |
| increase SWB particle count from K=1 to K=10 | SWB planning world model · 3D Food Chase | average total reward: 8.69 ± 9.81 -> 10.88 ± 9.90 | Helped |
| increase SWB particle count from K=10 to K=30 | SWB planning world model · 3D Food Chase | average total reward: 10.88 ± 9.90 -> 12.13 ± 8.59 | Helped |
| omit file-interaction summary during file-slot matching (ablation) | SWB · 2D Branching Sprites tracking | misses: 0.27 -> 17.3 | Hurt |
| omit file-interaction summary during file-slot matching (ablation) | SWB · 2D Branching Sprites tracking | position log-likelihood: -53.28 -> -83.92 | Hurt |
| update files directly from matched slots without glimpse encoding (ablation) | SWB · 2D Branching Sprites tracking | position log-likelihood: -53.28 -> -54.26 | No clear effect |
| update files directly from matched slots without glimpse encoding (ablation) | SWB · 3D Food Chase model training | convergence stability | Hurt |
Closest papers by meaning
- Efficient Exploration and Discriminative World Model Learning with an Object-Centric Abstraction (ICLR 2025 · similarity 0.88)
- Learning POMDP World Models from Observations with Language-Model Priors (NeurIPS 2026 · similarity 0.88)
- Learning Interactive World Model for Object-Centric Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Provable Representation with Efficient Planning for Partially Observable Reinforcement Learning (ICML 2024 · similarity 0.86)
- How to Explore with Belief: State Entropy Maximization in POMDPs (ICML 2024 · similarity 0.86)
- Dyn-O: Building Structured World Models with Object-Centric Representations (NeurIPS 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.