BriefGPT.xyz
Oct, 2014
高维数据的最大信息层次表示
Maximally Informative Hierarchical Representations of High-Dimensional Data
HTML
PDF
Greg Ver Steeg, Aram Galstyan
TL;DR
本研究介绍了一种基于概率函数的输入变量表示方法，并提出如何量化各层对原始数据信息提取的贡献，并应用于深度表示的无监督学习，具有可实施性和科学性。
Abstract
We consider a set of
probabilistic functions
of some input variables as a representation of the inputs. We present bounds on how informative a representation is about input data. We extend these bounds to
hierarchical r
→