智能体AI中涌现故障的定位：通过反事实回放恢复最小修复族
Localizing Emergent Failures in Agentic AI: Recovering Minimal Repair Families via Counterfactual Replay
浏览论文内容
中文总结 AI 辅助
本研究针对智能体AI中LLM智能体交互引发的涌现故障，提出GCJR方法恢复最小修复族，在基准和试点案例中实现1.000族精确匹配，显著减少回放或模型调用次数。
中文摘要 AI 辅助
智能体AI系统的故障可由多个大语言模型（LLM）智能体间交换消息的相互作用引发。逐点归因无法区分联合必要修复与替代单例修复。我们提出最小修复族恢复（MRFR）：恢复所有包含式最小事件集，其反事实回放可在声明的规模边界内恢复任务成功。我们提出图约束联合回放（GCJR），它从执行依赖图中切取与故障相关的事件，构建图可行的单例及对例候选，并通过与成对干净对应体回放进行验证。对于固定的回放结果，GCJR在其声明的图域内是精确的。在来自120个有向无环图（DAG）控制基准的90个范围内案例中，GCJR实现了1.000的族精确匹配，且与穷举搜索相比，将平均回放调用次数从56.3减少至25.3（降低55.1%）；在24个案例的四智能体LLM试点中，其再次实现1.000的族精确匹配，并将平均模型调用次数从21.0减少至10.0（降低52.4%）；单事件回放会遗漏联合必要修复。
英文摘要
Failures in agentic AI systems can arise from interactions among messages exchanged by multiple large language model (LLM) agents. Pointwise attribution cannot distinguish a jointly necessary repair from alternative singleton repairs. We formulate Minimal Repair Family Recovery (MRFR): recovering all inclusion-minimal event sets whose counterfactual replay restores task success within a declared size bound. We propose Graph-Constrained Joint Replay (GCJR), which slices failure-relevant events from an execution dependency graph, constructs graph-feasible singleton and pair candidates, and verifies them by replay with paired clean counterparts. For fixed replay outcomes, GCJR is exact within its declared graph domain. On 90 in-scope cases from a 120-DAG controlled benchmark, GCJR achieves 1.000 Family Exact Match while reducing mean replay calls from 56.3 to 25.3 (55.1%) relative to exhaustive search. On a 24-case, four-agent LLM pilot, it again achieves 1.000 Family Exact Match and reduces mean model calls from 21.0 to 10.0 (52.4%); single-event replay misses jointly necessary repairs.
发表机构
- Northeastern University（东北大学）
- Aalborg University（奥尔堡大学）
- Zhejiang University（浙江大学）
机构由 AI 辅助整理，请以论文原文为准。
