1.论文“A fast learning algorithm for deep belief nets”的“explaining away”现象的解释:
2.论文“A fast learning algorithm for deep belief nets”的整个过程及其“Complementary priors”的解释:
见:paper:A fast learning algorithm for deep belief nets和 [20140410] Complementary Prior
深度学习--深度信念网络(Deep Belief Network)、DBN的理解
代码
http://www.cs.toronto.edu/~hinton/MatlabForSciencePaper.html
运行mnistclassify.m
实验结果