TimePro: Efficient Multivariate Long-term Time Series Forecasting with Variable- and Time-Aware Hyper-state

TL;DR

TimePro introduces a novel hyper-state model for long-term multivariate time series forecasting, effectively capturing variable relationships and temporal features to improve accuracy with linear complexity.

Contribution

It proposes a variate- and time-aware hyper-state mechanism that preserves fine-grained temporal features and adaptively focuses on salient time points, advancing long-term forecasting methods.

Findings

Performs competitively on eight real-world benchmarks.

Maintains linear complexity in long-term forecasting.

Effectively captures variable relationships and temporal features.

Abstract

In long-term time series forecasting, different variables often influence the target variable over distinct time intervals, a challenge known as the multi-delay issue. Traditional models typically process all variables or time points uniformly, which limits their ability to capture complex variable relationships and obtain non-trivial time representations. To address this issue, we propose TimePro, an innovative Mamba-based model that constructs variate- and time-aware hyper-states. Unlike conventional approaches that merely transfer plain states across variable or time dimensions, TimePro preserves the fine-grained temporal features of each variate token and adaptively selects the focused time points to tune the plain state. The reconstructed hyper-state can perceive both variable relationships and salient temporal information, which helps the model make accurate forecasting. In experiments, TimePro performs competitively on eight real-world long-term forecasting benchmarks with satisfactory linear complexity. Code is available at https://github.com/xwmaxwma/TimePro.