BLNNdevs/BLNN: Bayesian Learning for Artificial Neural Networks

Software to build and train two layer feed-forward artificial neural networks using bayesian methods for weight training. The software impliments the No-U-Turn (NUTS) sampler and Hamiltonian Monte Carlo with dual averaging. Estimation of hyperparameter values can be done through the evidence procedure. This package can be used for multiple statistical models including, linear/multiple regression, logistic regression, classification, survival analysis, among others.

Getting started

Package details

Package repositoryView on GitHub
Installation Install the latest version of this package by entering the following in R:
BLNNdevs/BLNN documentation built on Dec. 10, 2019, 3:31 a.m.