VISTA: Unsupervised 2D Temporal Dependency Representations for Time Series Anomaly Detection

2025-04-03

AI Summary

In unsupervised time-series anomaly detection (TSAD), two key bottlenecks persist: poor noise robustness and insufficient modeling of complex temporal dependencies using one-dimensional representations. To address these, we propose a training-free, end-to-end lightweight detection framework. Our method first maps raw univariate/multivariate time series into a two-dimensional self-attention correlation matrix—explicitly capturing long-range temporal dependencies without parameterized modeling. Second, it integrates STL decomposition with pre-trained feature-driven multivariate aggregation to enhance noise resilience and cross-dataset generalization. The framework requires no fine-tuning, incurs minimal memory overhead, and enables efficient deployment. Evaluated on five standard multivariate TSAD benchmarks—including SMD, SMAP, MSL, SWaT, and WADI—it consistently outperforms state-of-the-art methods, achieving new SOTA performance across all datasets.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Multiagent Systems: Adversarial Agents

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications

Abstract

Time Series Anomaly Detection (TSAD) is essential for uncovering rare and potentially harmful events in unlabeled time series data. Existing methods are highly dependent on clean, high-quality inputs, making them susceptible to noise and real-world imperfections. Additionally, intricate temporal relationships in time series data are often inadequately captured in traditional 1D representations, leading to suboptimal modeling of dependencies. We introduce VISTA, a training-free, unsupervised TSAD algorithm designed to overcome these challenges. VISTA features three core modules: 1) Time Series Decomposition using Seasonal and Trend Decomposition via Loess (STL) to decompose noisy time series into trend, seasonal, and residual components; 2) Temporal Self-Attention, which transforms 1D time series into 2D temporal correlation matrices for richer dependency modeling and anomaly detection; and 3) Multivariate Temporal Aggregation, which uses a pretrained feature extractor to integrate cross-variable information into a unified, memory-efficient representation. VISTA's training-free approach enables rapid deployment and easy hyperparameter tuning, making it suitable for industrial applications. It achieves state-of-the-art performance on five multivariate TSAD benchmarks.

Problem

• Detects anomalies in unlabeled noisy time series data
• Captures complex temporal relationships in 2D representations
• Overcomes dependency on clean inputs in existing methods

Innovation

• STL decomposition for noise reduction
• 2D temporal correlation matrices
• Pretrained feature extractor integration