Abstract: Coupled with the exponential expansion of the data and the high computational complexity of Restricted Boltzmann Machines,efficient computing of Restricted Boltzmann Machines has become an important issue．Based on the detailed analysis,the article introduced Hadoop platform into Restricted Boltzmann Machines,and proposed Restricted Boltzmann Machines recommendation algorithm on cloud platform．The algorithm solves the problem of data correlation with replication mechanism,and divides traditional Restricted Boltzmann Machines process into several Hadoop jobs which implements parallel computing．In the experiments,the comparative analysis between Hadoop platform implementation and the previous implementation draws the conclusion that the Hadoop platform improves Restricted Boltzmann Machines computation efficiently under conditions of large data sets.

Key words: Collaborative filtering,Restricted boltzmann machines,Parallel processing,Cloud computing,Hadoop