Unverified50% confidenceFactExact time

Anomaly detection focuses on deviations of metric patterns relative to historical baselines, and can identify abnormal behavioral patterns even when absolute values remain within the normal range.

1

Sources

50%

Confidence

Long-term

Relevance

9/19/2026

First Seen

Sources

RADAR: Catching Elusive Gray Failures with Anomaly Detection

Related Entities

Anomaly Detectionconcept

Related Claims

Unverified针对特定触发词的后门攻击可让模型在大多数情况下表现正常，仅在遇到特定输入时才输出预设错误信息，使常规评测基准难以捕捉65% similar Unverified记忆溯源图（Provenance Graph）可追踪某条记忆影响了哪些后续决策，从而实现级联撤销或影响评估，与数据血缘（Data Lineage）概念一脉相承64% similar Unverified评估分数依赖测试数据集，数据集随业务变化会导致评分漂移，使历史横向比较产生误导64% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/933578

API

curl https://kongchang.com/api/v1/knowledge/claims/933578

MCP

get_claim(id=933578)