Going Beyond Conventional OOD Detection

2024-11-16

arXiv.org

AI Summary

Deep learning models often misclassify out-of-distribution (OOD) samples as in-distribution (ID) ones when spurious correlations or fine-grained similarities exist, and existing methods struggle to simultaneously ensure robustness and fine-grained discriminability. This paper proposes ASCOOD, a unified framework that identifies invariant features via pixel-level attribution, synthesizes virtual outlier samples intrinsically—without external OOD data—and jointly optimizes ID classification accuracy and uncertainty calibration on these synthetic samples. ASCOOD establishes the first unified paradigm integrating spurious-correlation robustness, fine-grained discrimination, and standard OOD detection. It introduces an in-model, attribution-driven mechanism for virtual outlier generation. Evaluated on six benchmark datasets, ASCOOD consistently achieves state-of-the-art performance across all three tasks: spurious-correlation-robust classification, fine-grained OOD detection, and conventional OOD detection.

Technology Category

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

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• 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 critical to ensure the safe deployment of deep learning models in critical applications. Deep learning models can often misidentify OOD samples as in-distribution (ID) samples. This vulnerability worsens in the presence of spurious correlation in the training set. Likewise, in fine-grained classification settings, detection of fine-grained OOD samples becomes inherently challenging due to their high similarity to ID samples. However, current research on OOD detection has largely ignored these challenging scenarios, focusing instead on relatively easier (conventional) cases. In this work, we present a unified Approach to Spurious, fine-grained, and Conventional OOD Detection (ASCOOD). First, we propose synthesizing virtual outliers from ID data by approximating the destruction of invariant features. We identify invariant features with the pixel attribution method using the model being learned. This approach eliminates the burden of curating external OOD datasets. Then, we simultaneously incentivize ID classification and predictive uncertainty towards the virtual outliers leveraging standardized feature representation. Our approach effectively mitigates the impact of spurious correlations and encourages capturing fine-grained attributes. Extensive experiments across six datasets demonstrate the merit of ASCOOD in spurious, fine-grained, and conventional settings. The code is available at: https://github.com/sudarshanregmi/ASCOOD/

Problem

• Out-of-Distribution Detection
• Deep Learning Safety
• Misleading Associations

Innovation

• ASCOOD
• Anomaly Generation
• Uncertainty Preservation