A Distributed Implementation of Training the Restricted Boltzmann Machine

Abstract

In these recent years, deep learning technique becomes very important in the artificial intelligence research, especially in the machine learning field. Deep learning works well in different applications in machine learning such as image, speech, document processing, etc. Since deep learning is related to a lot of mathematical calculations. Some well-known mathematical model running in behind of it, so it is hard to get start as a novice. As most of deep architectures [1], such as, Deep Belief Network (DBN) [2], Deep Boltzmann Machine [3], stacked auto-encoder [4], are related to or based on the Restricted Boltzmann Machine (RBM) [5]. In this report, we are focus on the training process [6] and distributed implementation of training the Restricted Boltzmann Machine, also evaluating the performances of Restricted Boltzmann Machine in distributed environment.