Learning with Adaptive Prototype Manifolds for Out-of-Distribution Detection

2026-02-05

AI Summary

This work addresses the limitations of existing prototype-based out-of-distribution (OOD) detection methods, which often rely on static homogeneity assumptions and suffer from a disconnect between learning and inference, thereby failing to accurately model complex data manifolds. To overcome these challenges, the authors propose APEX, a two-stage framework that enhances OOD detection through feature manifold optimization. First, guided by the Minimum Description Length (MDL) principle, APEX constructs an Adaptive Prototype Manifold (APM) that automatically determines the optimal prototype complexity for each class. Second, it introduces a Posterior-Aware OOD Scoring (PAOS) mechanism to effectively bridge the gap between training and inference. Extensive experiments demonstrate that APEX significantly outperforms current state-of-the-art methods on benchmarks such as CIFAR-100.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Learning with Manifolds
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Out-of-distribution (OOD) detection is a critical task for the safe deployment of machine learning models in the real world. Existing prototype-based representation learning methods have demonstrated exceptional performance. Specifically, we identify two fundamental flaws that universally constrain these methods: the Static Homogeneity Assumption (fixed representational resources for all classes) and the Learning-Inference Disconnect (discarding rich prototype quality knowledge at inference). These flaws fundamentally limit the model's capacity and performance. To address these issues, we propose APEX (Adaptive Prototype for eXtensive OOD Detection), a novel OOD detection framework designed via a Two-Stage Repair process to optimize the learned feature manifold. APEX introduces two key innovations to address these respective flaws: (1) an Adaptive Prototype Manifold (APM), which leverages the Minimum Description Length (MDL) principle to automatically determine the optimal prototype complexity $K_c^*$ for each class, thereby fundamentally resolving prototype collision; and (2) a Posterior-Aware OOD Scoring (PAOS) mechanism, which quantifies prototype quality (cohesion and separation) to bridge the learning-inference disconnect. Comprehensive experiments on benchmarks such as CIFAR-100 validate the superiority of our method, where APEX achieves new state-of-the-art performance.

Problem

• Out-of-distribution detection
• Prototype-based representation learning
• Static Homogeneity Assumption
• Learning-Inference Disconnect

Innovation

• Adaptive Prototype Manifold
• Posterior-Aware OOD Scoring
• Minimum Description Length
• Prototype Collision
• Out-of-Distribution Detection