HyBDM: Multi-Scale Hybrid Experts for Time Series Forecasting with Bidirectional Dependency Modeling
Machine Learning
Summary
The authors propose HyBDM, a new model for forecasting complex time series data that combines two parts: one focuses on long-term global patterns while the other captures short-term local changes. They designed special components for each part, such as bidirectional convolutions for global trends and a transformer for local details, and a system to merge their outputs effectively. Tests on six datasets show their model predicts more accurately and runs more efficiently than existing methods. This approach helps better understand both broad and detailed signals in multivariate time series data.
Authors
Wenqiang Ma, Chen Cheng, Xue Cheng, Jiarui Ye
Abstract
Time series forecasting (TSF) is vital to many applications, yet existing models often struggle to capture the heterogeneous long-range global patterns and short-range local variations in multivariate time series. While some approaches partially model these dependencies, they often do not jointly exploit temporal and feature-wise information. To address this challenge, we propose HyBDM, a multi-scale hybrid model that decomposes temporal dynamics into global patterns and local variations, which are modeled by two specialized experts. The Global Patterns Expert employs an enhanced BiConv-Mamba module that integrates bidirectional convolutions, an M-SSM layer, a forgetting mechanism, and a GDD-MLP module for cross-channel modeling. The Local Variations Expert uses a Local Window Transformer (LWT) to perform efficient locality-aware attention with reduced computational complexity. In addition, a Multi-Scale Patcher and a Long-Short Router enable multi-resolution representations and adaptive fusion of the two experts. Experiments on six benchmark datasets show that HyBDM outperforms state-of-the-art methods in both forecasting accuracy and computational efficiency, demonstrating its effectiveness in bridging global-local dependencies for multivariate TSF.