Hugging Face Papers ↗ · 2026-11-09
How Far Can Untrained Neural Networks Go in Anomaly Detection? From Tables to Pixels
媒体报道 Hugging Face Papers 原文语言:中文
原文摘录
论文摘要(作者报告,未经本站复测):Anomaly detection (AD) is often framed as a representation-learning problem, where strong performance is expected to require learned features or AD-specific optimization. However, this view can overlook the intrinsic separability of anomalies from normal references, even in the absence of learned representations.
以下是来源提供的简短摘录,完整内容请查看原文。