Disclaimer: This repository contains a reimplementation of the methods presented in this paper. This is not the original code used in the experimental section of the paper.
For a high-level overview of the approach refer to the following blog post.
The bal
package contains a Pytorch implementation of the Bayesian adaptive size and
skip connection layers, a custom implementation of a truncated normal distribution by
subclassing torch.Distribution
, and some other utilities.
The examples
folder contains notebooks and utility functions to visually demonstrate
the approach described in the paper and the blog post mentioned above.
@InProceedings{pmlr-v89-dikov19a,
title = {Bayesian Learning of Neural Network Architectures},
author = {Dikov, Georgi and Bayer, Justin},
booktitle = {Proceedings of Machine Learning Research},
pages = {730--738},
year = {2019},
volume = {89},
month = {16--18 Apr}
}