Correcting Deviations from Normality: A Reformulated Diffusion Model for Multi-Class Unsupervised Anomaly Detection

2025-03-25

AI Summary

In multi-class unsupervised anomaly detection, diffusion models struggle to simultaneously preserve structural fidelity and accurately reconstruct anomalous regions. To address this, we propose the Bias-Correction Diffusion Model (BCDM), the first method to explicitly model anomalies as class-agnostic, structured noise in the latent space. BCDM introduces an anomaly-aware noise scheduling scheme and a region-selective reverse process, enabling denoising exclusively within anomalous regions while preserving normal structures. Coupled with reconstruction-residual guidance and a class-agnostic anomaly discrimination mechanism, BCDM departs from conventional global denoising paradigms, achieving high-fidelity reconstruction and pixel-level anomaly localization. Extensive experiments on standard benchmarks demonstrate consistent improvements: pixel-level AUPRC increases by 11–14% over state-of-the-art methods. The source code is publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Recent advances in diffusion models have spurred research into their application for Reconstruction-based unsupervised anomaly detection. However, these methods may struggle with maintaining structural integrity and recovering the anomaly-free content of abnormal regions, especially in multi-class scenarios. Furthermore, diffusion models are inherently designed to generate images from pure noise and struggle to selectively alter anomalous regions of an image while preserving normal ones. This leads to potential degradation of normal regions during reconstruction, hampering the effectiveness of anomaly detection. This paper introduces a reformulation of the standard diffusion model geared toward selective region alteration, allowing the accurate identification of anomalies. By modeling anomalies as noise in the latent space, our proposed extbf{Deviation correction diffusion} (Ours) model preserves the normal regions and encourages transformations exclusively on anomalous areas. This selective approach enhances the reconstruction quality, facilitating effective unsupervised detection and localization of anomaly regions. Comprehensive evaluations demonstrate the superiority of our method in accurately identifying and localizing anomalies in complex images, with pixel-level AUPRC improvements of 11-14% over state-of-the-art models on well known anomaly detection datasets. The code is available at https://github.com/farzad-bz/DeCo-Diff

Problem

• Correcting structural integrity loss in multi-class anomaly detection
• Selectively altering anomalous regions without degrading normal areas
• Improving unsupervised anomaly localization accuracy in complex images

Innovation

• Reformulated diffusion model for selective region alteration
• Models anomalies as noise in latent space
• Enhances reconstruction quality for anomaly detection