Generalized Few-Shot Out-of-Distribution Detection

2025-08-07

AI Summary

To address weak out-of-distribution (OOD) detection generalization and inconsistent cross-scenario performance in few-shot open-world settings, this paper proposes GOOD, a generalized OOD detection framework. Methodologically, GOOD introduces the GS-balance theory—the first theoretical characterization of how generic knowledge guides few-shot OOD detection by tightening the upper bound on generalization error. It further designs an adaptive Knowledge Dynamic Embedding (KDE) mechanism that jointly leverages a Generic Knowledge Model (GKM) and Bayesian confidence estimation to achieve generalized-confidence-driven output distribution alignment. Evaluated across multiple real-world OOD benchmarks, GOOD consistently achieves superior detection accuracy and enhanced cross-domain consistency, demonstrating strong generalization capability, robustness to distribution shifts, and scalability to diverse downstream tasks and domains.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Few-shot Out-of-Distribution (OOD) detection has emerged as a critical research direction in machine learning for practical deployment. Most existing Few-shot OOD detection methods suffer from insufficient generalization capability for the open world. Due to the few-shot learning paradigm, the OOD detection ability is often overfit to the limited training data itself, thus degrading the performance on generalized data and performing inconsistently across different scenarios. To address this challenge, we proposed a Generalized Few-shot OOD Detection (GOOD) framework, which empowers the general knowledge of the OOD detection model with an auxiliary General Knowledge Model (GKM), instead of directly learning from few-shot data. We proceed to reveal the few-shot OOD detection from a generalization perspective and theoretically derive the Generality-Specificity balance (GS-balance) for OOD detection, which provably reduces the upper bound of generalization error with a general knowledge model. Accordingly, we propose a Knowledge Dynamic Embedding (KDE) mechanism to adaptively modulate the guidance of general knowledge. KDE dynamically aligns the output distributions of the OOD detection model to the general knowledge model based on the Generalized Belief (G-Belief) of GKM, thereby boosting the GS-balance. Experiments on real-world OOD benchmarks demonstrate our superiority. Codes will be available.

Problem

• Addressing insufficient generalization in few-shot OOD detection
• Reducing overfitting to limited training data in OOD detection
• Improving consistency across scenarios for few-shot OOD models

Innovation

• Auxiliary General Knowledge Model for OOD detection
• Knowledge Dynamic Embedding mechanism adaptively modulates guidance
• Generalized Belief aligns distributions to boost GS-balance