PaAno+: Multiscale Encoding and Cross-Variable Attention for Time Series Anomaly Detection

2026-06-18

AI Summary

This work addresses the longstanding trade-off between computational efficiency and modeling capacity in time series anomaly detection. The authors propose a lightweight model based on chunked representation learning that jointly captures hierarchical temporal patterns and complex inter-variable dependencies through a multi-receptive-field convolutional backbone, multi-scale adaptive attention aggregation, and an explicit cross-variable fusion mechanism. To enhance feature discriminability and generalization, the model incorporates a temporal chunk ordering pretext task and triplet loss during pretraining. Evaluated on the TSB-AD benchmark, the method achieves state-of-the-art accuracy in both univariate and multivariate settings—significantly outperforming PaAno in metrics such as VUS-PR—while maintaining low computational overhead, making it well-suited for real-time inference in resource-constrained environments.

Abstract

Time-series anomaly detection has significant practical value for industrial and medical monitoring, as well as other critical domains. Current Transformer- and large-model-based detection approaches incur excessive computational overhead, while existing lightweight alternatives are constrained by insufficient feature extraction and inadequate modeling of dependencies across multivariate variables. To mitigate the above drawbacks, this study develops a lightweight, efficient anomaly detection model, dubbed PaAno, within the patch-oriented representation learning paradigm. In the encoder module, a multiscale feature-extraction backbone is constructed using convolutional kernels with differentiated receptive fields to capture hierarchical temporal characteristics; subsequent cross-scale adaptive attention aggregation, combined with residual connection optimization, further stabilizes feature representation learning. A cross-variable fusion attention module is embedded to explicitly characterize inter-variable correlations, empowering the model to identify anomalous patterns amid intricate operational conditions. Moreover, a novel pretext task based on temporal patch-window sorting is customized to uncover intrinsic structural properties of time series, and triplet loss is leveraged to optimize the patch embedding space for enhanced feature discrimination. Extensive experiments on the TSB-AD benchmark demonstrate that the proposed PaAno achieves state-of-the-art detection accuracy on both univariate and multivariate tasks, yielding significant performance gains across evaluation metrics, including VUS-PR, relative to the original PaAno. Leveraging a compact network design, the presented model achieves favorable computational efficiency, enabling deployment on resource-limited terminals for real-time anomaly inference.

Problem

• time series anomaly detection
• multivariate dependencies
• lightweight models
• feature extraction
• computational overhead

Innovation

• multiscale encoding
• cross-variable attention
• patch-oriented representation
• temporal pretext task
• lightweight anomaly detection