Visually Debugging Restricted Boltzmann Machine Training with a 3D Example
Jason Yosinski, Hod Lipson · 2012
Restricted Boltzmann Machines (RBMs) are being applied to a growing number of prob-lems with great success. In the process of training an RBM one must pick a number of parameters, but often these parameters are brittle and produce poor performance when slightly off. Here we describe several use-ful visualizations to assist in choosing ap-propriate values for these parameters. We also demonstrate a successful application of an RBM to a unique domain: learning a rep-resentation of synthetic 3D shapes. 1.