Spline Activation

Spline activation functions are being explored as replacements for traditional activation functions in neural networks, aiming to improve model performance, interpretability, and training stability. Research focuses on integrating these functions into various architectures, including convolutional neural networks (CNNs), graph convolutional networks (GCNs), and multi-layer perceptrons (MLPs), with a particular emphasis on Kolmogorov-Arnold Networks (KANs) which utilize learnable spline-based activations. This research is driven by the need for more accurate and explainable models across diverse applications, from medical image analysis to predictive modeling in engineering, and continual learning.