Component-Based Out-of-Distribution Detection

2026-04-23

AI Summary

Existing methods struggle to detect out-of-distribution (OOD) samples composed of legitimate in-distribution components, facing an inherent trade-off between sensitivity to local anomalies and tolerance for in-distribution diversity. This work proposes the first training-free, component-level OOD detection framework, which decomposes inputs into functional components and separately evaluates local appearance shifts via a Component Shift Score (CSS) and inter-component compositional inconsistencies through a Compositional Consistency Score (CCS). By introducing a novel component-based detection paradigm, the method enables multi-granular anomaly discrimination and significantly outperforms existing approaches on both coarse- and fine-grained OOD detection tasks.

Technology Category

• Computer Vision: Object Detection & Categorization
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Responsible Web: Algorithmic accountability and transparency on the web

Abstract

Out-of-Distribution (OOD) detection requires sensitivity to subtle shifts without overreacting to natural In-Distribution (ID) diversity. However, from the viewpoint of detection granularity, global representation inevitably suppress local OOD cues, while patch-based methods are unstable due to entangled spurious-correlation and noise. And neither them is effective in detecting compositional OODs composed of valid ID components. Inspired by recognition-by-components theory, we present a training-free Component-Based OOD Detection (CoOD) framework that addresses the existing limitations by decomposing inputs into functional components. To instantiate CoOD, we derive Component Shift Score (CSS) to detect local appearance shifts, and Compositional Consistency Score (CCS) to identify cross-component compositional inconsistencies. Empirically, CoOD achieves consistent improvements on both coarse- and fine-grained OOD detection.

Problem

• Out-of-Distribution Detection
• Compositional OOD
• Detection Granularity
• Local OOD Cues
• In-Distribution Diversity

Innovation

• Component-Based OOD Detection
• Compositional OOD
• Training-Free Framework
• Component Shift Score
• Compositional Consistency Score