TSKANMixer: Kolmogorov-Arnold Networks with MLP-Mixer Model for Time Series Forecasting
This work addresses time series forecasting for fields like economics and healthcare, but it is incremental as it builds on existing TSMixer with a KAN layer.
The paper tackled time series forecasting by modifying the TSMixer architecture with a Kolmogorov-Arnold Network (KAN) layer, resulting in TSKANMixer, which improved prediction accuracy over the original TSMixer across multiple datasets and ranked among top-performing models.
Time series forecasting has long been a focus of research across diverse fields, including economics, energy, healthcare, and traffic management. Recent works have introduced innovative architectures for time series models, such as the Time-Series Mixer (TSMixer), which leverages multi-layer perceptrons (MLPs) to enhance prediction accuracy by effectively capturing both spatial and temporal dependencies within the data. In this paper, we investigate the capabilities of the Kolmogorov-Arnold Networks (KANs) for time-series forecasting by modifying TSMixer with a KAN layer (TSKANMixer). Experimental results demonstrate that TSKANMixer tends to improve prediction accuracy over the original TSMixer across multiple datasets, ranking among the top-performing models compared to other time series approaches. Our results show that the KANs are promising alternatives to improve the performance of time series forecasting by replacing or extending traditional MLPs.