Open Peer Review. Open Publishing. Open Access. Open Discussion. Open Directory. Open Recommendations. Open API. Open Source.
Bias-Variance Decomposition for Boltzmann Machines
Mahito Sugiyama, Koji Tsuda, Hiroyuki Nakahara
Feb 15, 2018 (modified: Feb 15, 2018)ICLR 2018 Conference Blind Submissionreaders: everyoneShow Bibtex
Abstract:We achieve bias-variance decomposition for Boltzmann machines using an information geometric formulation. Our decomposition leads to an interesting phenomenon that the variance does not necessarily increase when more parameters are included in Boltzmann machines, while the bias always decreases. Our result gives a theoretical evidence of the generalization ability of deep learning architectures because it provides the possibility of increasing the representation power with avoiding the variance inflation.
TL;DR:We achieve bias-variance decomposition for Boltzmann machines using an information geometric formulation.
Keywords:Boltzmann machine, bias-variance decomposition, information geometry
Enter your feedback below and we'll get back to you as soon as possible.