发表机构
School of Computer Science \& Statistics, Trinity College Dublin, College Green, Dublin 2, Ireland ADAPT Centre, Trinity College Dublin, College Green, Dublin 2, Ireland CKDelta, 28/29 Sir John Rogerson's Quay, Dublin 2, Ireland
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究多智能体分诊中分歧引发升级导致的相关一致性盲点问题,提出结合随机森林、k近邻智能体及校准元模型的ARAT系统。通过实验表明该系统能降低预测不足，跨数据集验证显示多样化产生有效分歧才提升安全，揭示相关故障被忽视的风险。
AI 中文摘要
分歧引发的升级会在多智能体仲裁中造成结构性盲点：随着基础学习器的改进，它们往往会趋同，削弱了对相关故障集中处的安全监控。我们将此称为相关一致性盲点，并提出了ARAT（用于警报分诊的仲裁推理智能体），这是一种定向星型系统，结合了归纳随机森林（RF）智能体、基于类比案例的k近邻（k-NN）智能体和一个校准元模型来减轻这种影响。在来自新南威尔士大学-网络入侵检测数据集的82332个保留样本上，57.2%的错误发生在一致性情况下，90.6%的危险预测不足即使在保守覆盖后也能避开基于分歧的监控；消融实验表明，强化基础学习器会增加错误相关性同时减少分歧。ARAT通过保守覆盖（降低2.6个百分点）和安全标志门（降低0.5个百分点）将预测不足相对于软投票从4.80%降低到1.70%，显示出架构上的优势。临床再入院的跨数据集验证支持了这些指标，表明只有当多样化产生有效的分歧而非趋同时才会提高安全性。这些结果表明，分歧引发的升级可能对相关故障视而不见，随着智能体管道部署越来越强大、相关的模型，这种风险可能会加剧。
英文摘要
Disagreement-triggered escalation can create a structural blind spot in multi-agent arbitration: as base learners improve, they tend to converge, weakening safety monitoring where correlated failures concentrate. We term this correlated agreement blindness and present ARAT (Arbitrated Reasoning Agents for Alarm Triage), a directed-star system combining an inductive Random Forest (RF) agent, an analogical case-based k-nearest neighbour (k-NN) agent, and a calibrated meta-model to mitigate this effect. On 82,332 holdout samples from the UNSW-NB15 network intrusion detection dataset, 57.2% of errors occur under agreement and 90.6% of dangerous under-predictions evade disagreement-based monitoring even after conservative override; ablation shows that strengthening base learners increases error correlation while reducing disagreement. ARAT reduces under-prediction relative to soft voting from 4.80% to 1.70% via conservative override (-2.6pp) and a safety-flag gate (-0.5pp), demonstrating architectural gains. Cross-dataset validation on clinical readmission supports these indicators, suggesting that diversification improves safety only when it generates productive disagreement rather than convergence. These results indicate that disagreement-triggered escalation can be blind to correlated failure, a risk that may intensify as agentic pipelines deploy increasingly capable, correlated models.
Comments14 pages, 2 figures, 3 tables. Accepted at PAAMS 2026; this is the author's pre-review submitted version
