TSRE: Channel-Aware Typical Set Refinement for Out-of-Distribution Detection

2025-11-19

AI Summary

In open-world settings, out-of-distribution (OOD) detection suffers from erroneous inclusion of anomalous activations into the typical set, primarily because existing activation-based methods neglect channel-wise discriminability, activation magnitude, and distribution skewness. To address this, we propose a channel-aware typical-set refinement framework: (1) an initial typical set is constructed based on per-channel discriminability and activation magnitude; (2) a skewness-guided refinement strategy suppresses spurious activations induced by skewed activation distributions; and (3) an activation correction mechanism is integrated with energy scoring to yield robust OOD scores. The method is architecture-agnostic—compatible with diverse backbone networks and scoring functions—and achieves state-of-the-art OOD detection performance on ImageNet-1K and CIFAR-100. Moreover, it demonstrates strong cross-model generalization capability.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Out-of-Distribution (OOD) detection is a critical capability for ensuring the safe deployment of machine learning models in open-world environments, where unexpected or anomalous inputs can compromise model reliability and performance. Activation-based methods play a fundamental role in OOD detection by mitigating anomalous activations and enhancing the separation between in-distribution (ID) and OOD data. However, existing methods apply activation rectification while often overlooking channel's intrinsic characteristics and distributional skewness, which results in inaccurate typical set estimation. This discrepancy can lead to the improper inclusion of anomalous activations across channels. To address this limitation, we propose a typical set refinement method based on discriminability and activity, which rectifies activations into a channel-aware typical set. Furthermore, we introduce a skewness-based refinement to mitigate distributional bias in typical set estimation. Finally, we leverage the rectified activations to compute the energy score for OOD detection. Experiments on the ImageNet-1K and CIFAR-100 benchmarks demonstrate that our method achieves state-of-the-art performance and generalizes effectively across backbones and score functions.

Problem

• Improves OOD detection by addressing channel characteristic oversight
• Rectifies activation sets using channel-aware discriminability analysis
• Mitigates distributional skewness in typical set estimation

Innovation

• Channel-aware typical set refinement for OOD detection
• Skewness-based refinement to mitigate distributional bias
• Rectified activations compute energy score for detection