A Systematic Comparison of Training Objectives for Out-of-Distribution Detection in Image Classification

TL;DR

This paper systematically compares four training objectives for out-of-distribution detection in image classification, revealing that Cross-Entropy Loss offers the most consistent OOD detection performance across datasets.

Contribution

It provides a comprehensive analysis of how different training objectives affect OOD detection, highlighting the effectiveness of Cross-Entropy Loss in various scenarios.

Findings

Cross-Entropy Loss performs best in overall OOD detection.

Prototype Loss and AP Loss are competitive in certain settings.

All objectives achieve similar in-distribution accuracy.

Abstract

Out-of-distribution (OOD) detection is critical in safety-sensitive applications. While this challenge has been addressed from various perspectives, the influence of training objectives on OOD behavior remains comparatively underexplored. In this paper, we present a systematic comparison of four widely used training objectives: Cross-Entropy Loss, Prototype Loss, Triplet Loss, and Average Precision (AP) Loss, spanning probabilistic, prototype-based, metric-learning, and ranking-based supervision, for OOD detection in image classification under standardized OpenOOD protocols. Across CIFAR-10/100 and ImageNet-200, we find that Cross-Entropy Loss, Prototype Loss, and AP Loss achieve comparable in-distribution accuracy, while Cross-Entropy Loss provides the most consistent near- and far-OOD performance overall; the other objectives can be competitive in specific settings.