Diffusion based Semantic Outlier Generation via Nuisance Awareness for Out-of-Distribution Detection

2024-08-27

arXiv.org

AI Summary

To address the challenge in near-distribution out-of-distribution (OOD) detection—where synthesized OOD samples exhibit excessive pixel-space distance from in-distribution (ID) data and thus fail to model fine-grained semantic shifts—this paper proposes a diffusion-based method for generating challenging OOD samples. The core innovation is the Semantic-Interference Dual-path guidance (SONA Guidance), which decouples control over semantic mismatch and noise-feature preservation directly in pixel space: it ensures generated OOD samples exhibit substantial semantic deviation from ID data while retaining highly similar low-level statistical characteristics. The method integrates semantic segmentation guidance, interference-region modeling, and end-to-end joint training with OOD detection. Evaluated on near-distribution OOD benchmarks, it achieves an AUROC of 88.0%, outperforming the prior state-of-the-art by approximately six percentage points, thereby significantly enhancing model sensitivity to subtle semantic shifts.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Generation

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Out-of-distribution (OOD) detection, which determines whether a given sample is part of the in-distribution (ID), has recently shown promising results through training with synthetic OOD datasets. Nonetheless, existing methods often produce outliers that are considerably distant from the ID, showing limited efficacy for capturing subtle distinctions between ID and OOD. To address these issues, we propose a novel framework, Semantic Outlier generation via Nuisance Awareness (SONA), which notably produces challenging outliers by directly leveraging pixel-space ID samples through diffusion models. Our approach incorporates SONA guidance, providing separate control over semantic and nuisance regions of ID samples. Thereby, the generated outliers achieve two crucial properties: (i) they present explicit semantic-discrepant information, while (ii) maintaining various levels of nuisance resemblance with ID. Furthermore, the improved OOD detector training with SONA outliers facilitates learning with a focus on semantic distinctions. Extensive experiments demonstrate the effectiveness of our framework, achieving an impressive AUROC of 88% on near-OOD datasets, which surpasses the performance of baseline methods by a significant margin of approximately 6%.

Problem

• Generating challenging semantic outliers for OOD detection
• Controlling semantic and nuisance regions in ID samples
• Improving OOD detector training with semantic-focused outliers

Innovation

• Uses diffusion models for outlier generation
• Separates control over semantic and nuisance regions
• Enhances OOD detection with semantic-focused training