Revitalizing Canonical Pre-Alignment for Irregular Multivariate Time Series Forecasting

2025-08-03

AI Summary

To address two key bottlenecks in irregular multivariate time series (IMTS) forecasting—sequence inflation and high computational cost induced by conventional Canonical Pre-Alignment (CPA), and the inability of graph-based models to effectively capture global variable dependencies—this paper proposes KAFNet. Built upon the CPA framework, KAFNet integrates Temporal Kernel Aggregation (TKA) and Frequency-domain Linear Attention (FLA), augmented by Pre-Convolution smoothing and a learnable sequence compression module. This design enables efficient global dependency modeling while maintaining low computational complexity. Extensive experiments on multiple IMTS benchmarks demonstrate that KAFNet achieves state-of-the-art (SOTA) forecasting accuracy, reduces model parameters by 7.2×, and accelerates both training and inference by 8.4× compared to existing methods.

Technology Category

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

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Irregular multivariate time series (IMTS), characterized by uneven sampling and inter-variate asynchrony, fuel many forecasting applications yet remain challenging to model efficiently. Canonical Pre-Alignment (CPA) has been widely adopted in IMTS modeling by padding zeros at every global timestamp, thereby alleviating inter-variate asynchrony and unifying the series length, but its dense zero-padding inflates the pre-aligned series length, especially when numerous variates are present, causing prohibitive compute overhead. Recent graph-based models with patching strategies sidestep CPA, but their local message passing struggles to capture global inter-variate correlations. Therefore, we posit that CPA should be retained, with the pre-aligned series properly handled by the model, enabling it to outperform state-of-the-art graph-based baselines that sidestep CPA. Technically, we propose KAFNet, a compact architecture grounded in CPA for IMTS forecasting that couples (1) Pre-Convolution module for sequence smoothing and sparsity mitigation, (2) Temporal Kernel Aggregation module for learnable compression and modeling of intra-series irregularity, and (3) Frequency Linear Attention blocks for the low-cost inter-series correlations modeling in the frequency domain. Experiments on multiple IMTS datasets show that KAFNet achieves state-of-the-art forecasting performance, with a 7.2$ imes$ parameter reduction and a 8.4$ imes$ training-inference acceleration.

Problem

• Modeling irregular multivariate time series with uneven sampling
• Reducing compute overhead from dense zero-padding in CPA
• Capturing global inter-variate correlations effectively

Innovation

• Pre-Convolution module for sequence smoothing
• Temporal Kernel Aggregation for irregularity modeling
• Frequency Linear Attention for inter-series correlations