Catalyst: Out-of-Distribution Detection via Elastic Scaling

2026-02-02

AI Summary

This work addresses the limited discriminability of existing out-of-distribution (OOD) detection methods by proposing Catalyst, a novel framework that leverages channel-wise statistics—such as mean, standard deviation, and maximum activation—from feature maps prior to global average pooling to dynamically generate input-dependent, adaptive scaling factors. These factors modulate widely used OOD scoring functions, including Energy, ReAct, and KNN, without requiring any architectural modifications, thereby enabling seamless integration into existing post-processing pipelines. Extensive experiments demonstrate that Catalyst significantly outperforms state-of-the-art approaches, reducing the average false positive rate by 32.87%, 27.94%, and 22.25% on CIFAR-10, CIFAR-100, and ImageNet, respectively.

Technology Category

• Computer Vision: Object Detection & Categorization
• Machine Learning: Calibration & Uncertainty Quantification
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• 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
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Out-of-distribution (OOD) detection is critical for the safe deployment of deep neural networks. State-of-the-art post-hoc methods typically derive OOD scores from the output logits or penultimate feature vector obtained via global average pooling (GAP). We contend that this exclusive reliance on the logit or feature vector discards a rich, complementary signal: the raw channel-wise statistics of the pre-pooling feature map lost in GAP. In this paper, we introduce Catalyst, a post-hoc framework that exploits these under-explored signals. Catalyst computes an input-dependent scaling factor ($\gamma$) on-the-fly from these raw statistics (e.g., mean, standard deviation, and maximum activation). This $\gamma$ is then fused with the existing baseline score, multiplicatively modulating it -- an ``elastic scaling''-- to push the ID and OOD distributions further apart. We demonstrate Catalyst is a generalizable framework: it seamlessly integrates with logit-based methods (e.g., Energy, ReAct, SCALE) and also provides a significant boost to distance-based detectors like KNN. As a result, Catalyst achieves substantial and consistent performance gains, reducing the average False Positive Rate by 32.87 on CIFAR-10 (ResNet-18), 27.94% on CIFAR-100 (ResNet-18), and 22.25% on ImageNet (ResNet-50). Our results highlight the untapped potential of pre-pooling statistics and demonstrate that Catalyst is complementary to existing OOD detection approaches.

Problem

• Out-of-Distribution detection
• deep neural networks
• safety
• OOD detection
• distribution shift

Innovation

• out-of-distribution detection
• elastic scaling
• pre-pooling statistics
• post-hoc framework
• feature map statistics