An LLM-driven generate-test-feedback loop with a multi-scale mutation strategy finds more and more diverse failure scenarios for black-box decision-making policies than MDPFuzz and random testing.
Towards Fault Tolerance in Multi-Agent Reinforcement Learning
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Agent faults pose a significant threat to the performance of multi-agent reinforcement learning (MARL) algorithms, introducing two key challenges. First, agents often struggle to extract critical information from the chaotic state space created by unexpected faults. Second, transitions recorded before and after faults in the replay buffer affect training unevenly, leading to a sample imbalance problem. To overcome these challenges, this paper enhances the fault tolerance of MARL by combining optimized model architecture with a tailored training data sampling strategy. Specifically, an attention mechanism is incorporated into the actor and critic networks to automatically detect faults and dynamically regulate the attention given to faulty agents. Additionally, a prioritization mechanism is introduced to selectively sample transitions critical to current training needs. To further support research in this area, we design and open-source a highly decoupled code platform for fault-tolerant MARL, aimed at improving the efficiency of studying related problems. Experimental results demonstrate the effectiveness of our method in handling various types of faults, faults occurring in any agent, and faults arising at random times.
citation-role summary
citation-polarity summary
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Exploring Critical Testing Scenarios for Decision-Making Policies: An LLM Approach
An LLM-driven generate-test-feedback loop with a multi-scale mutation strategy finds more and more diverse failure scenarios for black-box decision-making policies than MDPFuzz and random testing.