Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Semi-Supervised Generative Models for Multiagent Trajectories
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 |
|---|---|---|---|
| add ground-truth weak labels (supervised S-MAT) vs unsupervised U-MAT | MAT · NBA 10-step | avg L2: 2.31 -> 1.94 | Helped |
| add ground-truth weak labels (supervised S-MAT) vs unsupervised U-MAT | MAT · NBA 10-step | final L2: 4.80 -> 3.99 | Helped |
| use MAT variants instead of DAG-Net | MAT · NBA offense 40-step | avg L2: 8.98 -> 8.11 | Helped |
| use unsupervised U-MAT instead of DAG-Net | MAT · NBA offense 40-step | avg L2: 8.98 -> 9.01 | No clear effect |
| use unsupervised U-MAT instead of DAG-Net | MAT · NBA offense 40-step | final L2: 14.08 -> 13.28 | Helped |
| use S-MAT instead of DAG-Net | MAT · NBA defense 40-step | final L2: 9.76 -> 8.45 | Helped |
| add unlabeled data via semi-supervised MAT instead of S-MAT | MAT · Soccer 10-step | avg L2: 2.87 -> 2.81 | Helped |
| use full-covariance MAT instead of diagonal MAT-DIAG | MAT · Soccer 10-step | avg L2: 2.84 -> 2.81 | Helped |
| add graph interactions: GVRNN vs VRNN | VRNN · Soccer 10-step | avg L2: 2.91 -> 2.89 | No clear effect |
| add unlabeled data via MAT instead of S-MAT | MAT · Soccer event detection | accuracy: 0.82 -> 0.88 | Helped |
| add unlabeled data via MAT instead of S-MAT | MAT · Soccer event detection | F1: 0.85 -> 0.92 | Helped |
| use inter-agent modeling (MA-RNN vs MA-RNN-DIAG) | MA-RNN · Soccer event detection | accuracy: 0.74 -> 0.85 | Helped |
Closest papers by meaning
- Diverse Generation for Multi-Agent Sports Games (CVPR 2019 · similarity 0.90)
- Sports-Traj: A Unified Trajectory Generation Model for Multi-Agent Movement in Sports (ICLR 2025 · similarity 0.88)
- Unified Uncertainty-Aware Diffusion for Multi-Agent Trajectory Modeling (CVPR 2025 · similarity 0.86)
- Event2Tracking: Reconstructing Multi-Agent Soccer Trajectories Using Long-Term Multimodal Context (AAAI 2025 · similarity 0.86)
- Trajectory-Class-Aware Multi-Agent Reinforcement Learning (ICLR 2025 · similarity 0.86)
- A Variational Autoencoder for Neural Temporal Point Processes with Dynamic Latent Graphs (AAAI 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.