#neural architecture

Linear-Time B-splines KAN Architecture Reduces Computational Cost

New approach accelerates Kolmogorov-Arnold Networks while maintaining expressibility and interpretability.