@article{Oh_Adamczewski_Park_2020, title={Radial and Directional Posteriors for Bayesian Deep Learning}, volume={34}, url={https://ojs.aaai.org/index.php/AAAI/article/view/5976}, DOI={10.1609/aaai.v34i04.5976}, abstractNote={<p>We propose a new variational family for Bayesian neural networks. We decompose the variational posterior into two components, where the <em>radial</em> component captures the strength of each neuron in terms of its magnitude; while the <em>directional</em> component captures the statistical dependencies among the weight parameters. The dependencies learned via the directional density provide better modeling performance compared to the widely-used Gaussian mean-field-type variational family. In addition, the strength of input and output neurons learned via our posterior provides a structured way to compress neural networks. Indeed, experiments show that our variational family improves predictive performance and yields compressed networks simultaneously.</p>}, number={04}, journal={Proceedings of the AAAI Conference on Artificial Intelligence}, author={Oh, Changyong and Adamczewski, Kamil and Park, Mijung}, year={2020}, month={Apr.}, pages={5298-5305} }