MTS-JEPA: Multi-Resolution Joint-Embedding Predictive Architecture for Time-Series Anomaly Prediction

2026-02-04

AI Summary

Multivariate time series anomaly forecasting is often hindered by representation collapse and the difficulty of capturing precursor signals across diverse temporal scales. This work proposes a Multi-Resolution Joint Embedding Predictive Architecture (JEPA) that explicitly disentangles transient shocks from long-term trends through a soft codebook bottleneck and incorporates multi-resolution prediction targets as an intrinsic regularizer to effectively prevent degenerate solutions. By integrating multi-scale modeling, soft quantization, and contrastive learning, the method substantially enhances model stability and representational capacity. Evaluated on standard benchmarks, the approach achieves state-of-the-art early-warning performance, significantly improving anomaly prediction accuracy while effectively mitigating representation collapse.

Technology Category

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

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Multivariate time series underpin modern critical infrastructure, making the prediction of anomalies a vital necessity for proactive risk mitigation. While Joint-Embedding Predictive Architectures (JEPA) offer a promising framework for modeling the latent evolution of these systems, their application is hindered by representation collapse and an inability to capture precursor signals across varying temporal scales. To address these limitations, we propose MTS-JEPA, a specialized architecture that integrates a multi-resolution predictive objective with a soft codebook bottleneck. This design explicitly decouples transient shocks from long-term trends, and utilizes the codebook to capture discrete regime transitions. Notably, we find this constraint also acts as an intrinsic regularizer to ensure optimization stability. Empirical evaluations on standard benchmarks confirm that our approach effectively prevents degenerate solutions and achieves state-of-the-art performance under the early-warning protocol.

Problem

• time-series anomaly prediction
• representation collapse
• multi-resolution
• precursor signals
• Joint-Embedding Predictive Architecture

Innovation

• multi-resolution
• soft codebook bottleneck
• representation collapse
• regime transitions
• early-warning anomaly prediction