Research map / Reinforcement learning / MDPs and planning
Collective Multiagent Sequential Decision Making Under Uncertainty
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 closed loop EM policy instead of SMFU | CDec-MDP EM · Robots moving to a goal | normalized solution quality: more than 20% higher | Helped |
| use open loop EM instead of SMFU | CDec-MDP EM · Robots moving to a goal | normalized solution quality: about 5%-10% higher | Helped |
| use closed loop policy instead of open loop policy at convergence | CDec-MDP EM · Robots 5x5 grid | normalized reward | Helped |
| use closed loop policy instead of open loop in early iterations | CDec-MDP EM · Robots 5x5 grid | normalized reward | Hurt |
| increase iterations for EM vs SMFU/FP-SAP | CDec-MDP EM · Robots 5x5 grid | iterations to converge: 20 -> 200 | Hurt |
| use EM instead of MIP solver | CDec-MDP EM · Robots grid size | scalability: 7 x 7 | Helped |
| increase plan horizon (fixed 5x5 grid) | EM variants vs previous approaches · Robots 5x5 grid | normalized reward | No clear effect |
| use EM instead of average-approximation methods SMFU/FP-SAP | CDec-MDP EM · Taxi supply-demand, Singapore | average profit per taxi per day | Helped |
| increase MaxVar of demand stochasticity from 1 to 10 | SMFU and FP-SAP · Taxi supply-demand | profit relative to EM: 1 to 10 | Hurt |
| increase MaxVar from 1 to 10 | average approximation accuracy obj/obj* · Taxi supply-demand | obj/obj*: 1 to 10 | Hurt |
| evaluate average-flow objective vs true sampled objective | SMFU, FP-SAP, MIP · Robots moving to a goal | obj/obj* | Hurt |
Closest papers by meaning
- Credit Assignment For Collective Multiagent RL With Global Rewards (NeurIPS 2018 · similarity 0.91)
- Factored Online Planning in Many-Agent POMDPs (AAAI 2024 · similarity 0.87)
- Locally Interdependent Multi-Agent MDP: Theoretical Framework for Decentralized Agents with Dynamic Dependencies (ICML 2024 · similarity 0.87)
- Multiagent Gumbel MuZero: Efficient Planning in Combinatorial Action Spaces (AAAI 2024 · similarity 0.86)
- Learning Decentralized Partially Observable Mean Field Control for Artificial Collective Behavior (ICLR 2024 · similarity 0.86)
- Graph Diffusion for Robust Multi-Agent Coordination (ICML 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.