FINE-GRAIN INFERENCE ON OUT-OF-DISTRIBUTION DATA WITH HIERARCHICAL CLASSIFICATION

Aug 22, 2026

Machine learning systems must make trustworthy decisions under out-of-distribution (OOD) inputs, yet many methods reduce OOD handling to a binary decision based on confidence. Binary anomaly detection becomes uninformative when OOD samples overlap strongly with training data. This work introduces a hierarchical OOD detection model that predicts at multiple granularity levels: as inputs grow more ambiguous, predictions become coarser and more conservative, enabling explainable hierarchy-aware diagnoses. Experiments validate the approach for both fine- and coarse-grained OOD settings.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages22

Reading Time55 min

AccessFree to download

Abstract

Introduction

Limitations of binary OOD detection

Motivation for fine-grained OOD

Hierarchical classification approach

Path-wise probabilities and inference stopping criteria

Hierarchical OOD metrics and uncertainty-aware predictions

Hierarchical loss for fine-grained OOD

FINE-GRAIN INFERENCE ON OUT-OF-DISTRIBUTION DATA WITH HIERARCHICAL CLASSIFICATION Randolph Linderman 1 Jingyang Zhang 1 Nathan Inkawhich2 Hai Li 1 Yiran Chen 1 1 Department of Electrical and Computer Engineering Duke University Durham, NC 27708 { [first}.{last}@duke.edu](first}.{last}@duke.edu) 2Information Directorate Air Force Research Laboratory Rome, NY 13441 [nathan.inkawhich@us.af.mil](nathan.inkawhich@us.af.mil) ABSTRACT Machine learning methods must be trusted to make appropriate decisions in real-world environments, even when faced with out-of-distribution (OOD) samples. Many current approaches simply aim to detect OOD examples and alert the user when an unrecognized input is given. However, when the OOD sample significantly overlaps with the training data, a binary anomaly detection is not interpretable or explainable, and provides little information to the user. We propose a new model for OOD detection that makes predictions at varying levels of granularity—as the inputs become more ambiguous, the model predictions become coarser and more conservative. Consider an animal classifier that encounters an unknown bird species and a car. Both cases are OOD, but the user gains more information if the classifier recognizes that its uncertainty over the particular species is too large and predicts “bird” instead of detecting it as OOD. Furthermore, we diagnose the classifier’s performance at each level of the hierarchy improving the explainability and interpretability of the model’s predictions. We demonstrate the effectiveness of hierarchical classifiers for both fine-and coarse-grained OOD tasks. The code is available at [https://github](https://github) .com/rwl93/ hierarchical-ood. 1 INTRODUCTION Real-world computer vision systems will encounter out-of-distribution (OOD) samples while making or informing consequential decisions. Therefore, it is crucial to design machine learning methods that make reasonable predictions for anomalous inputs that are outside the scope of the training distribution. Recently, research has focused on detecting inputs during inference that are OOD for the training distribution (Ahmed & Courville, 2020 ; Hendrycks & Gimpel, 2017 ; Hendrycks et al., 2019 ; Hsu et al., 2020 ; Huang & Li, 2021 ; Lakshminarayanan et al., 2017 ; Lee et al., 2018 ; Liang et al., 2018 ; Liu et al., 2020 ; Neal et al., 2018 ; Roady et al., 2020 ; Inkawhich et al., 2022) . These methods typically use a threshold on the model’s “confidence” to produce a binary decision indicating if the sample is in-distribution (ID) or OOD. However, binary decisions based on model heuristics offer little interpretability or explainability. The fundamental problem is that there are many ways for a sample to be out-of-distribution. Ideally, a model should provide more nuanced information about how a sample differs from the training data. For example, if a bird classifier is presented with a novel bird species, we would like it to recognize that the sample is a bird rather than simply reporting OOD. On the contrary, if the bird classifier is shown an MNIST digit then it should indicate that the digit is outside its domain of expertise. Recent studies have shown that fine-grained OOD samples are significantly more difficult to detect, especially when there is a large number of training classes (Ahmed & Courville, 2020 ; Huang & Li, 2021 ; Roady et al., 2020 ; Zhang et al., 2023 ; Inkawhich et al., 2021) . We argue that the difficulty stems from trying to address two opposing objectives: learning semantically meaningful features to discriminate between ID classes while also maintaining tight decision boundaries to avoid misclassification on fine-grain OOD samples (Ahmed & Courville, 2020 ; Huang & Li, 2021) . We hypothesize that additional information about the relationships between classes could help determine those decision boundaries and simultaneously offer more interpretable predictions. To address these chal

FAQs

Why is binary out-of-distribution detection often insufficient?

Binary decisions based on confidence thresholds provide limited interpretability, and they offer little useful information when OOD inputs significantly overlap with training data.

How does the hierarchical model improve OOD predictions?

It predicts at varying granularity levels in a class hierarchy, producing coarser and more conservative outputs as uncertainty increases, so users gain more informative and actionable explanations.

What is the role of the hierarchical loss function in this work?

A new loss function is proposed to better handle the fine-grained OOD scenario within the hierarchical softmax classification framework.

Aug 22, 2026

Upload by Bill Black