FreSH: Frequency-Segmented Hierarchical Multi-Expert Framework for Multivariate Time Series Classification

2026-08-08

AI Summary

This work addresses the challenge of simultaneously achieving effective multi-scale modeling, handling class imbalance, and maintaining computational efficiency in multivariate time series classification. To this end, we propose a frequency-partitioned hierarchical mixture-of-experts framework that leverages adaptive multi-scale time–frequency analysis to enable localized specialization while preserving global contextual awareness. A dynamic fusion mechanism is introduced to emphasize discriminative features, and a robust optimization objective enhances model stability and generalization under class-imbalanced conditions. Evaluated on 30 UEA benchmark datasets and real-world vibration data, our method achieves state-of-the-art classification accuracy with a more compact model size and superior computational efficiency.

Abstract

Multivariate Time Series Classification (MTSC) demands models that can effectively capture complex temporal patterns across multiple scales while remaining computationally efficient. However, existing approaches generally struggle to reconcile fine-grained representation learning, especially under class imbalance and real-world constraints. In this paper, we present FreSH, a Frequency-Segmented Hierarchical Multi-Expert Framework designed to address these challenges. FreSH introduces a new perspective for MTSC by enabling adaptive, multi-scale analysis of temporal signals, allowing different aspects of the data to be modeled in a complementary and coordinated manner. By combining localized specialization with holistic context modeling, FreSH achieves strong representational capacity without incurring excessive computational overhead. An adaptive fusion strategy further enhances flexibility, enabling the model to dynamically emphasize the most informative components of the input. In addition, we incorporate a more robust optimization objective that improves learning stability across varying sample difficulties and class distributions. Extensive evaluations on 30 UEA benchmark datasets and real-world vibration data demonstrate that FreSH consistently outperforms state-of-the-art methods in classification accuracy, while substantially reducing model size and efficiency.

Problem

• Multivariate Time Series Classification
• class imbalance
• multi-scale temporal patterns
• computational efficiency
• fine-grained representation learning

Innovation

• Frequency-Segmented
• Hierarchical Multi-Expert
• Multivariate Time Series Classification
• Adaptive Fusion
• Robust Optimization