Research map / Multi-agent systems and games / Cooperative multi-agent RL
InterAgent: Physics-based Multi-agent Command Execution via Diffusion on Interaction Graphs
Bin Li, Ruichi Zhang, Han Liang, Jingyan Zhang, Juze Zhang, Xin Chen, Lan Xu, Jingyi Yu, Jingya Wang
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 relative-state exteroception with fully connected interaction graph (ablation) | InterAgent 3-stream · InterHuman test | FID: 0.676 -> 0.634 | Improved |
| add sparse edge attention to the interaction graph (FIG to SIG) (ablation) | InterAgent 3-stream · InterHuman test | FID: 0.634 -> 0.582 | Improved |
| use 3 streams instead of 1 stream in DiT block (ablation) | InterAgent with FIG · InterHuman test | FID: 0.828 -> 0.634 | Improved |
| use 3 streams instead of 2 streams in DiT block (ablation) | InterAgent with FIG · InterHuman test | FID: 0.662 -> 0.634 | Improved |
| use 2 streams instead of 1 stream in DiT block (ablation) | InterAgent with FIG · InterHuman test | R-precision Top-3: 0.523 -> 0.608 | Improved |
| edge-based sparse attention at ratio 1/2 instead of non-sparse (ablation) | InterAgent · InterHuman test | FID: 0.634 -> 0.582 | Improved |
| use a very high sparsity (ratio 1/8) edge-based attention (ablation) | InterAgent · InterHuman test | FID: 0.634 -> 0.643 | No clear effect |
| use joint-based instead of edge-based sparsity at ratio 1/2 (ablation) | InterAgent · InterHuman test | FID: 0.582 -> 0.619 | Hurt |
| train end-to-end physics-based diffusion over interaction graphs instead of generate-then-track (InterGen++) | physics-based multi-agent control · InterHuman test | FID: 0.943 -> 0.582 | Improved |
| use InterAgent instead of InterGen++ | Multimodality metric · InterHuman test | MModality: 2.482 -> 1.903 | Hurt |
Closest papers by meaning
- CooHOI: Learning Cooperative Human-Object Interaction with Manipulated Object Dynamics (NeurIPS 2024 · similarity 0.87)
- LLawCo: Learning Laws of Cooperation for Modeling Embodied Multi-Agent Behavior (ICML 2026 · similarity 0.86)
- Scaling Large Language Model-based Multi-Agent Collaboration (ICLR 2025 · similarity 0.86)
- Emergent Coordination in Multi-Agent Language Models (ICLR 2026 · similarity 0.86)
- AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors (ICLR 2024 · similarity 0.86)
- G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems (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.
