arXiv：cs.AI· Boxuan Zhang, jianing Zhu, Xiaotian Han, Ruixiang Tang·· 18 小时前
Dream2Discern：用硬隐变体预测多智能体系统运行时故障
Dream2Discern: Learning from Hard Latent Variants for Runtime Failure Prediction in Multi-Agent Systems
AI 导读
Dream2Discern 是一种运行时审计器，通过自身预测损失引导的隐式增强训练：Dream 阶段扰动表示算子生成难分类的硬隐变体，Discern 阶段用原始标签从变体预测执行结果，变体随审计器学习不断再生，无需额外采集轨迹或验证结果。在 AFTraj-2K 和 StepShield 上，该方法的轨迹级加权准确率均超过 90%，优于同类故障审计器，并兼顾预测准确性与及时性。
正文
Abstract:Reliable deployment of LLM-based multi-agent systems calls for runtime auditors that predict terminal outcomes from observed execution prefixes. These auditors typically learn from a limited corpus of trajectories with verified outcomes. Expanding this corpus requires costly execution and verification, yet broader failure coverage alone provides no explicit criterion for identifying prefixes that the current auditor finds difficult to classify. To this end, we propose Dream2Discern, a runtime auditor trained through implicit augmentation guided by its own prediction loss. During Dream, we select perturbations of the auditor's representation operators to increase prediction loss on recorded prefixes, yielding hard latent variants. During Discern, the auditor learns to predict execution outcomes from these variants using the original labels. The variants are regenerated as the auditor learns, turning a limited corpus into a renewable source of hard examples and requiring no additional trajectory collection or outcome verification. Extensive evaluations on AFTraj-2K and StepShield show that Dream2Discern outperforms competitive failure auditors and achieves over 90% trajectory-level weighted accuracy on both benchmarks while balancing prediction accuracy and timeliness. Project page: this https URL
| Comments: | 27 pages, 6 figures |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2610.16071 [cs.LG] |
| (or arXiv:2610.16071v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.16071 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Boxuan Zhang [view email]
[v1]
Mon, 12 Oct 2026 02:19:44 UTC (2,014 KB)
来源：arXiv：cs.AI · arxiv.org