Temporal Query Network for Efficient Multivariate Time Series Forecasting

2025-05-19

AI Summary

To address insufficient modeling of cross-variable dependencies in multivariate time series forecasting (MTSF), this paper proposes the Temporal Query (TQ) mechanism: learnable, periodically shifted query vectors capture global inter-variable patterns, while key-value pairs derived from raw data encode local sample-wise dependencies—enabling, for the first time, decoupled modeling of global patterns and local associations. Built upon TQ, we design TQNet—a lightweight, single-layer architecture combining attention and MLP—that achieves a breakthrough trade-off between accuracy and efficiency. Evaluated on 12 real-world datasets, TQNet consistently outperforms state-of-the-art methods, establishing new accuracy benchmarks. Its computational complexity scales nearly linearly with sequence length and dimensionality, offering exceptional efficiency and robustness—particularly in high-dimensional settings—without sacrificing predictive performance.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Planning, Routing, and Scheduling: Temporal Planning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Sufficiently modeling the correlations among variables (aka channels) is crucial for achieving accurate multivariate time series forecasting (MTSF). In this paper, we propose a novel technique called Temporal Query (TQ) to more effectively capture multivariate correlations, thereby improving model performance in MTSF tasks. Technically, the TQ technique employs periodically shifted learnable vectors as queries in the attention mechanism to capture global inter-variable patterns, while the keys and values are derived from the raw input data to encode local, sample-level correlations. Building upon the TQ technique, we develop a simple yet efficient model named Temporal Query Network (TQNet), which employs only a single-layer attention mechanism and a lightweight multi-layer perceptron (MLP). Extensive experiments demonstrate that TQNet learns more robust multivariate correlations, achieving state-of-the-art forecasting accuracy across 12 challenging real-world datasets. Furthermore, TQNet achieves high efficiency comparable to linear-based methods even on high-dimensional datasets, balancing performance and computational cost. The code is available at: https://github.com/ACAT-SCUT/TQNet.

Problem

• Improving multivariate time series forecasting accuracy
• Capturing global and local inter-variable correlations
• Balancing model performance and computational efficiency

Innovation

• Temporal Query technique captures global inter-variable patterns
• Single-layer attention and lightweight MLP ensure efficiency
• State-of-the-art accuracy on 12 real-world datasets