Balanced Hyperbolic Embeddings Are Natural Out-of-Distribution Detectors

TL;DR

This paper demonstrates that balanced hyperbolic embeddings serve as effective out-of-distribution detectors, outperforming existing methods across multiple datasets and scoring functions by leveraging hierarchical hyperbolic space.

Contribution

Introduces Balanced Hyperbolic Learning, a novel hyperbolic class embedding algorithm that optimizes hierarchical structure and balance for improved OOD detection.

Findings

Hyperbolic embeddings outperform existing OOD detection methods.

The approach beats other hyperbolic and contrastive methods.

Embeddings enable hierarchical OOD generalization.

Abstract

Out-of-distribution recognition forms an important and well-studied problem in deep learning, with the goal to filter out samples that do not belong to the distribution on which a network has been trained. The conclusion of this paper is simple: a good hierarchical hyperbolic embedding is preferred for discriminating in- and out-of-distribution samples. We introduce Balanced Hyperbolic Learning. We outline a hyperbolic class embedding algorithm that jointly optimizes for hierarchical distortion and balancing between shallow and wide subhierarchies. We then use the class embeddings as hyperbolic prototypes for classification on in-distribution data. We outline how to generalize existing out-of-distribution scoring functions to operate with hyperbolic prototypes. Empirical evaluations across 13 datasets and 13 scoring functions show that our hyperbolic embeddings outperform existing out-of-distribution approaches when trained on the same data with the same backbones. We also show that our hyperbolic embeddings outperform other hyperbolic approaches, beat state-of-the-art contrastive methods, and natively enable hierarchical out-of-distribution generalization.