用多样性衡量新奇性，无需密度建模即可高效检测分布外样本。
Vendi Novelty Scores for Out-of-Distribution Detection
- 基于相似性多样性度量，从数据集整体和类别内双重角度评估新奇性。
- 在多个图像分类任务上达到当前最佳性能，仅用1%训练数据仍有效。
- 线性时间复杂度，适合内存受限或数据访问受限场景使用。
分布外（OOD）检测对机器学习系统安全部署至关重要。现有后处理检测器通常依赖模型置信度或特征空间中的似然估计，常受分布假设限制。本文提出第三种范式，从多样性视角重新定义OOD检测。我们引入Vendi新奇性评分（VNS），基于Vendi得分（VS）——一类基于相似性的多样性度量。VNS通过量化测试样本使分布内特征集的VS提升程度，提供无需密度建模的合理新奇性判断。VNS具备线性时间复杂度、非参数化特性，并自然融合类别条件（局部）与数据集级（全局）的新奇信号。在多个图像分类基准和网络架构上，VNS均达到当前最优性能。尤为显著的是，仅使用1%训练数据计算的VNS仍保持良好表现，适用于内存或数据访问受限场景。
原文摘要 · Abstract (English)
Out-of-distribution (OOD) detection is critical for the safe deployment of machine learning systems. Existing post-hoc detectors typically rely on model confidence scores or likelihood estimates in feature space, often under restrictive distributional assumptions. In this work, we introduce a third paradigm and formulate OOD detection from a diversity perspective. We propose the Vendi Novelty Score (VNS), an OOD detector based on the Vendi Scores (VS), a family of similarity-based diversity metrics. VNS quantifies how much a test sample increases the VS of the in-distribution feature set, providing a principled notion of novelty that does not require density modeling. VNS is linear-time, non-parametric, and naturally combines class-conditional (local) and dataset-level (global) novelty signals. Across multiple image classification benchmarks and network architectures, VNS achieves state-of-the-art OOD detection performance. Remarkably, VNS retains this performance when computed using only 1% of the training data, enabling deployment in memory- or access-constrained settings.
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。