Time series forecasting with Hahn Kolmogorov-Arnold networks

ArXi:2601.18837v2 Announce Type: replace Recent Transformer- and MLP-based models have nstrated strong performance in long-term time series forecasting, yet Transformers remain limited by their quadratic complexity and permutation-equivariant attention, while MLPs exhibit spectral bias. We propose HaKAN, a versatile model based on Kolmogoro-Arnold Networks (KANs), leveraging Hahn polynomial-based learnable activation functions and providing a lightweight and interpretable alternative for multivariate time series forecasting.