Selective Denoising Diffusion Model for Time Series Anomaly Detection

2026-02-27

AI Summary

This work proposes AnomalyFilter, a novel diffusion-based approach for time series anomaly detection that addresses the challenge of accurately reconstructing normal patterns—a key limitation in existing methods. AnomalyFilter introduces, for the first time, a selective denoising mechanism into the diffusion framework: during training, Gaussian noise is masked to focus learning on anomalous deviations, while at inference, the model directly denoises the original input, selectively correcting only the anomalous components and preserving the intact normal structure. This design effectively constructs a dedicated filtering mechanism tailored for anomaly detection. Evaluated on five standard benchmarks, AnomalyFilter significantly reduces reconstruction error on normal samples and consistently outperforms state-of-the-art methods in detection performance, demonstrating both its effectiveness and methodological innovation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Time-Series/Data Streams

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Content-based information diffusion
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Time series anomaly detection (TSAD) has been an important area of research for decades, with reconstruction-based methods, mostly based on generative models, gaining popularity and demonstrating success. Diffusion models have recently attracted attention due to their advanced generative capabilities. Existing diffusion-based methods for TSAD rely on a conditional strategy, which reconstructs input instances from white noise with the aid of the conditioner. However, this poses challenges in accurately reconstructing the normal parts, resulting in suboptimal detection performance. In response, we propose a novel diffusion-based method, named AnomalyFilter, which acts as a selective filter that only denoises anomaly parts in the instance while retaining normal parts. To build such a filter, we mask Gaussian noise during the training phase and conduct the denoising process without adding noise to the instances. The synergy of the two simple components greatly enhances the performance of naive diffusion models. Extensive experiments on five datasets demonstrate that AnomalyFilter achieves notably low reconstruction error on normal parts, providing empirical support for its effectiveness in anomaly detection. AnomalyFilter represents a pioneering approach that focuses on the noise design of diffusion models specifically tailored for TSAD.

Problem

• time series anomaly detection
• diffusion models
• reconstruction error
• anomaly detection
• denoising

Innovation

• selective denoising
• diffusion model
• time series anomaly detection
• AnomalyFilter
• noise masking