Robust Functional Basis Neural Network Method With the Normalization
A robust loss function tailored to handle outliers effectively applied within a neural network based on basis expansion is introduced, and a novel functional normalization technique that normalizes the input function values of hidden layers with respect to time points is proposed.