Supplementary Material for " Nonlinear Sufficient Dimension Reduction with a Stochastic Neural Network "

Mar-27-2025, 12:14:03 GMT–Neural Information Processing Systems 

For the full data case, we have ς = 0, i.e., the gradient