用于视觉异常值检测的双曲度量学习
Hyperbolic Metric Learning for Visual Outlier Detection
AI总结：
针对视觉数据层次结构在欧式空间中难以捕捉的问题，提出基于双曲几何的OOD检测度量框架，实验证明其在CIFAR-10和CIFAR-100上显著降低了FPR95。
摘要：
分布外（OOD）检测在安全关键应用中部署深度学习模型至关重要。然而，视觉数据固有的层次概念结构对OOD检测很有帮助，却常常被基于欧几里得几何的传统方法糟糕地捕捉。本研究提出一个度量框架，利用双曲几何的优势进行OOD检测。受先前用合成异常值细化OOD数据决策边界的工作启发，我们将该方法扩展到双曲空间。有趣的是，我们发现合成异常值在双曲空间中并不像在欧几里得空间中那样有利于OOD检测。此外，我们探讨了OOD检测性能与双曲嵌入维度之间的关系，以解决资源受限环境中的实际问题。大量实验表明，与欧几里得方法相比，我们的框架将CIFAR-10和CIFAR-100上OOD检测的FPR95分别从22%降至15%和从49%降至28%。
英文摘要：
Out-Of-Distribution (OOD) detection is critical to deploy deep learning models in safety-critical applications. However, the inherent hierarchical concept structure of visual data, which is instrumental to OOD detection, is often poorly captured by conventional methods based on Euclidean geometry. This work proposes a metric framework that leverages the strengths of Hyperbolic geometry for OOD detection. Inspired by previous works that refine the decision boundary for OOD data with synthetic outliers, we extend this method to Hyperbolic space. Interestingly, we find that synthetic outliers do not benefit OOD detection in Hyperbolic space as they do in Euclidean space. Furthermore we explore the relationship between OOD detection performance and Hyperbolic embedding dimension, addressing practical concerns in resource-constrained environments. Extensive experiments show that our framework improves the FPR95 for OOD detection from 22\% to 15\% and from 49% to 28% on CIFAR-10 and CIFAR-100 respectively compared to Euclidean methods.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
