Multivariate Time-series Anomaly Detection via Dynamic Model Pool & Ensembling

2026-01-05

arXiv.org

AI Summary

This work addresses key limitations in existing multivariate time series anomaly detection methods, which often suffer from sensitivity to model selection, rigid ensemble strategies, and poor scalability with increasing dimensionality. To overcome these challenges, the authors propose DMPEAD, a novel framework that introduces a dynamic model pool specifically designed for multivariate time series. The framework maintains model diversity within the pool through parameter transfer and diversity-aware metrics, and enables adaptive updates via a meta-model guided by similarity-based strategies. Furthermore, it employs a proxy metric to rank candidate models and selects the top-k for ensemble construction, thereby enhancing detection performance. Extensive experiments on eight real-world datasets demonstrate that DMPEAD significantly outperforms current baselines, exhibiting both strong adaptability and superior scalability.

Technology Category

• Machine Learning: Ensemble Methods
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Search and Optimization: Metareasoning and Metaheuristics

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
• Web Mining and Content Analysis: Models for Web evolution
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Multivariate time-series (MTS) anomaly detection is critical in domains such as service monitor, IoT, and network security. While multi-model methods based on selection or ensembling outperform single-model ones, they still face limitations: (i) selection methods rely on a single chosen model and are sensitive to the strategy; (ii) ensembling methods often combine all models or are restricted to univariate data; and (iii) most methods depend on fixed data dimensionality, limiting scalability. To address these, we propose DMPEAD, a Dynamic Model Pool and Ensembling framework for MTS Anomaly Detection. The framework first (i) constructs a diverse model pool via parameter transfer and diversity metric, then (ii) updates it with a meta-model and similarity-based strategy for adaptive pool expansion, subset selection, and pool merging, finally (iii) ensembles top-ranked models through proxy metric ranking and top-k aggregation in the selected subset, outputting the final anomaly detection result. Extensive experiments on 8 real-world datasets show that our model outperforms all baselines, demonstrating superior adaptability and scalability.

Problem

• Multivariate Time-series
• Anomaly Detection
• Model Ensembling
• Scalability
• Model Selection

Innovation

• Dynamic Model Pool
• Ensemble Learning
• Multivariate Time-series Anomaly Detection
• Adaptive Model Selection
• Scalability