Research map / Multi-agent systems and games / Cooperative multi-agent RL
Learning Robust Multi-Agent Policies via Selective Adversarial Fault Induction
David Henry Mguni, Yaqi Sun, Haojun Chen, Wanrong Yang, Amir Darabi, Larry Olanrewaju Orimoloye, Yaodong Yang
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 |
|---|---|---|---|
| add MARTA switcher-adversary training | QMIX · Traffic Junction | final test return: -54.2 -> -48.2 | Improved |
| add MARTA training | QMIX · LBF-5x5-4p-1f | final test return: 0.65 -> 0.94 | Improved |
| add MARTA training | QMIX · SMACv2 3m | final test return: 1.14 -> 2.45 | Improved |
| add MARTA training | VDN · SMACv2 3m | final test return: 2.57 -> 5.57 | Improved |
| add MARTA training | VDN · LBF-5x5-4p-1f | final test return: 0.94 -> 0.97 | No clear effect |
| add MARTA training | VDN · Traffic Junction | collision/failure rate (%): 17.5 -> 5.0 | Improved |
| replace learned Switcher with Bernoulli random trigger (ablation) | MARTA · MPE SimpleTag | test return | Hurt |
| add MARTA to MADDPG backbone vs M3DDPG | MADDPG · MPE SimpleTag | test return | Improved |
| compare MARTA to EIR under aligned fixed fault (Case 1) | EIR · TJ, LBF, MPE, SMAC-3m | robustness metrics | No clear effect |
| compare MARTA to EIR under dynamic faults (Case 2) | EIR · TJ, LBF, MPE, SMAC-3m | robustness metrics | Improved |
| increase switch cost c (ablation) | MARTA · Traffic Junction | fault activations | Mixed |
| add MARTA under higher malfunction probability p (ablation) | QMIX · Traffic Junction | final return | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- QMIX — QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning (ICML 2018) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Empirical Study on Robustness and Resilience in Cooperative Multi-Agent Reinforcement Learning (NeurIPS 2025 · similarity 0.90)
- Byzantine Robust Cooperative Multi-Agent Reinforcement Learning as a Bayesian Game (ICLR 2024 · similarity 0.90)
- AdvEvo-MARL: Shaping Internalized Safety through Adversarial Co-Evolution in Multi-Agent Reinforcement Learning (ICML 2026 · similarity 0.90)
- Robust Multi-Agent Reinforcement Learning with Stochastic Adversary (ICML 2025 · similarity 0.88)
- Robust and Diverse Multi-Agent Learning via Rational Policy Gradient (NeurIPS 2025 · similarity 0.88)
- Breaking the Curse of Multiagency in Robust Multi-Agent Reinforcement Learning (ICML 2025 · similarity 0.88)
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.
