TFKAN: Time-Frequency KAN for Long-Term Time Series Forecasting

2025-06-15

AI Summary

Existing long-term time series forecasting methods predominantly rely on time-domain Kolmogorov–Arnold Network (KAN) modeling, neglecting periodic and repetitive patterns inherently present in the frequency domain. To address this limitation, we propose TF-KAN—a novel time-frequency dual-domain decoupled KAN framework featuring a synergistic two-branch architecture. The time-domain branch employs adaptive KAN activations to capture nonlinear dynamics, while the frequency-domain branch leverages short-time Fourier transform (STFT), augmented with dimensionality lifting and selective upsampling, to enhance periodic feature extraction. Cross-domain feature interaction enables complementary fusion between branches. Extensive experiments on multiple benchmark datasets demonstrate that TF-KAN consistently outperforms state-of-the-art methods, substantiating the efficacy of time-frequency collaborative modeling in improving both forecasting accuracy and robustness.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Computer Vision: Multi-modal Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Kolmogorov-Arnold Networks (KANs) are highly effective in long-term time series forecasting due to their ability to efficiently represent nonlinear relationships and exhibit local plasticity. However, prior research on KANs has predominantly focused on the time domain, neglecting the potential of the frequency domain. The frequency domain of time series data reveals recurring patterns and periodic behaviors, which complement the temporal information captured in the time domain. To address this gap, we explore the application of KANs in the frequency domain for long-term time series forecasting. By leveraging KANs' adaptive activation functions and their comprehensive representation of signals in the frequency domain, we can more effectively learn global dependencies and periodic patterns. To integrate information from both time and frequency domains, we propose the $ extbf{T}$ime-$ extbf{F}$requency KAN (TFKAN). TFKAN employs a dual-branch architecture that independently processes features from each domain, ensuring that the distinct characteristics of each domain are fully utilized without interference. Additionally, to account for the heterogeneity between domains, we introduce a dimension-adjustment strategy that selectively upscales only in the frequency domain, enhancing efficiency while capturing richer frequency information. Experimental results demonstrate that TFKAN consistently outperforms state-of-the-art (SOTA) methods across multiple datasets. The code is available at https://github.com/LcWave/TFKAN.

Problem

• Bridging time and frequency domains in KANs for forecasting
• Enhancing global dependencies and periodic pattern learning
• Proposing dual-branch TFKAN to integrate domain-specific features

Innovation

• Dual-branch architecture for time-frequency processing
• Dimension-adjustment strategy in frequency domain
• Adaptive activation functions for global dependencies