End-To-End Self-tuning Self-supervised Time Series Anomaly Detection

2024-04-03

arXiv.org

AI Summary

Time-series anomaly detection (TSAD) faces challenges in unsupervised, general-purpose identification of diverse anomaly types—including spikes, abrupt level shifts, and trend deviations—while existing self-supervised methods exhibit high sensitivity to data augmentation strategies and lack mechanisms for automatic, label-free augmentation tuning. This paper proposes TSAP: the first end-to-end self-tuning self-supervised framework for TSAD. TSAP jointly optimizes discrete augmentation types and continuous hyperparameters via a differentiable augmentation architecture, guided by an unsupervised validation loss. By integrating contrastive learning with gradient-based hyperparameter optimization, TSAP automatically aligns augmentation policies with underlying anomaly patterns. Evaluated on multi-type anomaly benchmarks, TSAP significantly outperforms state-of-the-art self-supervised methods, achieving superior detection accuracy and cross-anomaly generalization.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Search and Optimization: Learning to Search

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 heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Time series anomaly detection (TSAD) finds many applications such as monitoring environmental sensors, industry KPIs, patient biomarkers, etc. A two-fold challenge for TSAD is a versatile and unsupervised model that can detect various different types of time series anomalies (spikes, discontinuities, trend shifts, etc.) without any labeled data. Modern neural networks have outstanding ability in modeling complex time series. Self-supervised models in particular tackle unsupervised TSAD by transforming the input via various augmentations to create pseudo anomalies for training. However, their performance is sensitive to the choice of augmentation, which is hard to choose in practice, while there exists no effort in the literature on data augmentation tuning for TSAD without labels. Our work aims to fill this gap. We introduce TSAP for TSA"on autoPilot", which can (self-)tune augmentation hyperparameters end-to-end. It stands on two key components: a differentiable augmentation architecture and an unsupervised validation loss to effectively assess the alignment between augmentation type and anomaly type. Case studies show TSAP's ability to effectively select the (discrete) augmentation type and associated (continuous) hyperparameters. In turn, it outperforms established baselines, including SOTA self-supervised models, on diverse TSAD tasks exhibiting different anomaly types.

Problem

• Develop self-tuning model for time series anomaly detection
• Automate augmentation selection without labeled data
• Improve detection of diverse anomaly types

Innovation

• End-to-end self-tuning augmentation hyperparameters
• Differentiable augmentation architecture design
• Unsupervised validation loss for alignment