PaAno: Patch-Based Representation Learning for Time-Series Anomaly Detection

2026-02-01

AI Summary

This work addresses the high computational cost, substantial memory consumption, and limited performance gains of existing large-model-based (e.g., Transformer) approaches for time series anomaly detection under rigorous evaluation. The authors propose PaAno, a lightweight patch-based method that segments time series into short patches, embeds them using a 1D convolutional network, and trains the model via triplet loss combined with pretext-task-based self-supervised learning. During inference, anomaly scores are derived from the similarity between the current patch’s embedding and learned normal patterns in the embedding space. By avoiding complex architectures, PaAno achieves state-of-the-art performance on the TSB-AD benchmark, significantly outperforming both lightweight and heavyweight baselines, and is well-suited for real-time and resource-constrained applications.

Technology Category

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

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Although recent studies on time-series anomaly detection have increasingly adopted ever-larger neural network architectures such as transformers and foundation models, they incur high computational costs and memory usage, making them impractical for real-time and resource-constrained scenarios. Moreover, they often fail to demonstrate significant performance gains over simpler methods under rigorous evaluation protocols. In this study, we propose Patch-based representation learning for time-series Anomaly detection (PaAno), a lightweight yet effective method for fast and efficient time-series anomaly detection. PaAno extracts short temporal patches from time-series training data and uses a 1D convolutional neural network to embed each patch into a vector representation. The model is trained using a combination of triplet loss and pretext loss to ensure the embeddings capture informative temporal patterns from input patches. During inference, the anomaly score at each time step is computed by comparing the embeddings of its surrounding patches to those of normal patches extracted from the training time-series. Evaluated on the TSB-AD benchmark, PaAno achieved state-of-the-art performance, significantly outperforming existing methods, including those based on heavy architectures, on both univariate and multivariate time-series anomaly detection across various range-wise and point-wise performance measures.

Problem

• time-series anomaly detection
• computational efficiency
• resource-constrained scenarios
• large neural networks
• performance evaluation

Innovation

• patch-based representation
• time-series anomaly detection
• lightweight architecture
• 1D convolutional neural network
• triplet loss