Revitalizing Canonical Pre-Alignment for Irregular Multivariate Time Series Forecasting

TL;DR

This paper introduces KAFNet, a novel model that revitalizes the use of Canonical Pre-Alignment in irregular multivariate time series forecasting, achieving state-of-the-art results with reduced complexity.

Contribution

KAFNet combines pre-alignment with innovative modules for smoothing, compression, and frequency domain correlation modeling, outperforming existing graph-based methods.

Findings

Achieves state-of-the-art forecasting accuracy.

Reduces model parameters by 7.2 times.

Speeds up training and inference by 8.4 times.

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$\times$ parameter reduction and a 8.4$\times$ training-inference acceleration.