Research map / Multi-agent systems and games / Cooperative multi-agent RL
IPOMDP-Net: A Deep Neural Network for Partially Observable Multi-Agent Planning Using Interactive POMDPs
Yanlin Han, Piotr Gmytrasiewicz
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 |
|---|---|---|---|
| replace model-free ADRQN with model-based IPOMDP-net | agent policy network · Tiger (fixed env) | average reward: 1.29 ± 0.35 -> 1.32 ± 0.29 | No clear effect |
| replace model-free ADRQN with model-based IPOMDP-net | agent policy network · UAV (fixed env) | average reward: 8.98 ± 0.97 -> 9.19 ± 0.72 | No clear effect |
| replace ADRQN with IPOMDP-net | agent policy network · Maze 4×4 | average reward: 0.12 ± 0.08 -> 0.18 ± 0.05 | Improved |
| replace ADRQN with IPOMDP-net | agent policy network · Maze 10×10 | average reward: -0.73 ± 0.17 -> -0.52 ± 0.10 | Improved |
| replace ADRQN with IPOMDP-net, transferred from 10×10 training | agent policy network · Maze 16×16 | average reward: -1.58 ± 0.39 -> -0.88 ± 0.21 | Improved |
| preassign IPOMDP-net weights with true model vs symbolic I-POMDP (QMDP) | IPOMDP-net · Tiger | average reward: 2.26 ± 0.09 -> 2.25 ± 0.11 | No clear effect |
| preassign IPOMDP-net weights with true model vs symbolic I-POMDP (QMDP) | IPOMDP-net · Maze 16×16 | average reward: -0.96 ± 0.09 -> -0.99 ± 0.10 | No clear effect |
| use SARSOP planner instead of QMDP in symbolic I-POMDP | I-POMDP · Maze 16×16 | average reward: -0.96 ± 0.09 -> -0.81 ± 0.10 | Improved |
| use SARSOP planner instead of QMDP in symbolic I-POMDP | I-POMDP · UAV | average reward: 9.09 ± 0.45 -> 9.27 ± 0.52 | No clear effect |
| increase map size from 4×4 to 16×16 | ADRQN · Maze | average reward: 0.12 ± 0.08 -> -1.58 ± 0.39 | Hurt |
| increase map size from 4×4 to 16×16 | IPOMDP-net trained · Maze | average reward: 0.18 ± 0.05 -> -0.88 ± 0.21 | Hurt |
Closest papers by meaning
- Learning Logic Specifications for Policy Guidance in POMDPs: an Inductive Logic Programming Approach (AAAI 2025 · similarity 0.87)
- Partially Observable Hierarchical Reinforcement Learning with AI Planning (Student Abstract) (AAAI 2024 · similarity 0.87)
- Multi-Environment POMDPs with Finite-Horizon Objectives (NeurIPS 2026 · similarity 0.86)
- Factored Online Planning in Many-Agent POMDPs (AAAI 2024 · similarity 0.86)
- Provable Representation with Efficient Planning for Partially Observable Reinforcement Learning (ICML 2024 · similarity 0.85)
- Belief-State Query Policies for User-Aligned POMDPs (NeurIPS 2024 · 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.
