Communication-based Multi-agent Reinforcement Learning (com-MARL) on SH Poisoned
36.6Average Winning RateBehaviorGuard
Evaluation results
| Method | Links | |
|---|---|---|
| BehaviorGuard2026.05 | 36.6 | |
| SHINE2026.05 | 35.8 | |
| BIRD2026.05 | 29.3 | |
| Direct retraining2026.05 | 27.8 | |
| Original2026.05 | 15.9 |