RoCA: Robust Contrastive One-class Time Series Anomaly Detection with Contaminated Data

2025-03-24

AI Summary

Time-series anomaly detection faces three practical challenges: (1) the oversimplified assumption of a single normal pattern induces bias; (2) partial assumptions violate fundamental anomaly detection (AD) principles; and (3) training data often contain latent anomalies, undermining model robustness. To address these, we propose the first unsupervised framework unifying one-class classification and contrastive learning. It estimates latent anomalies via a dynamic anomaly scoring mechanism and adaptively recalibrates the decision boundary accordingly—thereby jointly achieving multi-source normality modeling, strict adherence to AD principles, and robust training under contaminated data. Our core innovation is an anomaly-score-driven boundary self-calibration mechanism, inspired by the anomaly exposure paradigm. Evaluated on AIOps and multiple high-dimensional and univariate benchmark datasets, our method achieves average improvements of 5–10% over COCA and establishes new state-of-the-art performance.

Technology Category

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

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

The accumulation of time-series signals and the absence of labels make time-series Anomaly Detection (AD) a self-supervised task of deep learning. Methods based on normality assumptions face the following three limitations: (1) A single assumption could hardly characterize the whole normality or lead to some deviation. (2) Some assumptions may go against the principle of AD. (3) Their basic assumption is that the training data is uncontaminated (free of anomalies), which is unrealistic in practice, leading to a decline in robustness. This paper proposes a novel robust approach, RoCA, which is the first to address all of the above three challenges, as far as we are aware. It fuses the separated assumptions of one-class classification and contrastive learning in a single training process to characterize a more complete so-called normality. Additionally, it monitors the training data and computes a carefully designed anomaly score throughout the training process. This score helps identify latent anomalies, which are then used to define the classification boundary, inspired by the concept of outlier exposure. The performance on AIOps datasets improved by 6% compared to when contamination was not considered (COCA). On two large and high-dimensional multivariate datasets, the performance increased by 5% to 10%. RoCA achieves the highest average performance on both univariate and multivariate datasets. The source code is available at https://github.com/ruiking04/RoCA.

Problem

• Addresses limitations in time-series anomaly detection assumptions
• Improves robustness with contaminated training data handling
• Enhances performance on univariate and multivariate datasets

Innovation

• Fuses one-class classification and contrastive learning
• Monitors training data for latent anomalies
• Uses anomaly score to define classification boundary