提出新方法GEODE，让模型在未知的异常数据下仍能稳定检测。
GEODE: Angle-Adaptive OOD Detection with Universal Scorer Compatibility
- 通过角度自适应损失，模拟真实异常数据的特征几何结构。
- 在CIFAR-10上对7种评分器均表现优异，近域异常检测准确率89.0-92.3。
- 无需额外数据，兼容主流检测器，适合部署时未知异常场景。
Outlier Exposure（OE）是当前最强的基于训练的异常检测方法之一，但在不同评分器间存在性能权衡（如在MSP上强，而在KNN上弱），且需使用精心设计的辅助数据。本文揭示其原理：OE的特征位于与真实近似异常数据相同的几何位置，其中靠近边界的四分位数贡献了几乎全部性能提升。本质上，OE实现的是边界校准，而非异常覆盖。为此，提出GEODE（GEOmetry-preserving DEtection），通过一种角度自适应范数损失，使目标值按样本与最近类别均值的余弦相似度动态调整，从而在关键区域保留特征几何结构。四个基于神经坍缩理论的定理支持该设计。GEODE在七个标准评分器上均表现稳健，在CIFAR-10上近域异常检测的AUROC达89.0–92.3，远域达到93.05，且无任何评分器出现灾难性失效。相比原始交叉熵，在相同训练轮次下表现更优。结合OE后，于CIFAR-10上取得95.0（MSP）/94.8（KNN）的性能，超越OE在所有评分器上的表现；在CIFAR-100和WRN-28-10上也持续领先，提升达+4.5（能量评分，3种子）。不同于将异常推向分类器零空间的方法（如PFS，KNN AUROC仅14.38，低于随机），GEODE的自适应目标保留了距离类评分器依赖的几何结构。
原文摘要 · Abstract (English)
Outlier Exposure (OE) is among the strongest training-based OOD detectors on standard benchmarks but exhibits scorer-dependent tradeoffs (e.g., strong on MSP, weak on KNN) and requires curated auxiliary data. We show why OE works: its features sit at the same geometric locus as real near-OOD data, with the boundary-adjacent quartile driving nearly all of OE's gain. OE is boundary calibration, not OOD coverage. GEODE (GEOmetry-preserving DEtection) replicates this calibration synthetically through an angle-adaptive norm loss in which targets scale per-sample with cosine similarity to the nearest class mean, preserving feature geometry where boundary structure matters. Four theorems grounded in neural collapse justify the design. GEODE works across all seven standard scorers on CIFAR-10 (near-OOD AUROC 89.0-92.3, far-OOD reaching 93.05; no catastrophic failure on any scorer). Since the OOD regime is unknown at deployment, this is the test that matters. GEODE outperforms vanilla CE at matched epoch counts. Combined with OE, GEODE reaches 95.0 MSP / 94.8 KNN on CIFAR-10 and beats OE on every scorer on CIFAR-100. The gains hold on WRN-28-10 (+4.5 Energy, 3 seeds). Unlike methods that push OOD into the classifier null space (e.g., PFS, 14.38 KNN AUROC, worse than random), GEODE's adaptive target preserves the geometry that distance-based scorers depend on.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。