Research map / Architectures and efficiency / Pruning and efficient tuning
Solving Large-Scale Pursuit-Evasion Games Using Pre-trained Strategies
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 |
|---|---|---|---|
| pre-train the pursuer policy against varied evader strategies, then fine-tune it for each best response | PSRO pursuer best-response computation · 20*20 grid, T=10, 2 units | worst-case utility | Helped |
| pre-train the pursuer policy against varied evader strategies, then fine-tune it for each best response | PSRO pursuer best-response computation · 10*10, 20*20, and 50*50 grids | worst-case utility | Helped |
| pre-train and fine-tune the pursuer policy | PSRO · 100*100 grid, T=51, 8 units | scalability | Helped |
| pre-train and fine-tune the pursuer policy | PSRO · Singapore map, T=15, 6 units | worst-case utility | Helped |
| replace historical locations with current positions and remaining time in the state (ablation) | PPO-PSRO · 20*20 grid, T=10, 2 units | worst-case utility | No clear effect |
| add graph embeddings instead of representing vertices only by index (ablation) | PSRO · 20*20 grid, T=10, 2 units | worst-case utility | Helped |
| add exit-node information proximity to LINE graph embeddings (ablation) | PSRO with LINE embeddings · 20*20 grid, T=10, 2 units | worst-case utility | Helped |
| replace multi-task pre-training with meta-learning (ablation) | pre-trained pursuer policy in PSRO · 20*20 grid, T=10, 2 units | pre-training time | Hurt |
| use multi-task learning rather than meta-learning for pre-training (ablation) | pre-trained pursuer policy in PSRO · 20*20 grid, T=10, 2 units | fine-tuning performance | Helped |
Closest papers by meaning
- R2PS: Worst-Case Robust Real-Time Pursuit Strategies under Partial Observability (ICLR 2026 · similarity 0.87)
- Tree-Based Stochastic Optimization for Solving Large-Scale Urban Network Security Games (AAAI 2026 · similarity 0.87)
- Equilibrium Policy Generalization: A Reinforcement Learning Framework for Cross-Graph Zero-Shot Generalization in Pursuit-Evasion Games (NeurIPS 2025 · similarity 0.86)
- GUARD: Constructing Realistic Two-Player Matrix and Security Games for Benchmarking Game-Theoretic Algorithms (NeurIPS 2025 · similarity 0.84)
- Bench-MFG: A Benchmark Suite for Learning in Stationary Mean Field Games (NeurIPS 2026 · similarity 0.83)
- Learning Mean Field Games on Sparse Graphs: A Hybrid Graphex Approach (ICLR 2024 · similarity 0.83)
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.