Parallelized Training of Restricted Boltzmann Machines using Markov-Chain Monte Carlo Methods

TL;DR

This paper presents a distributed parallel training method for Restricted Boltzmann Machines using Horovod, significantly reducing training time and enhancing practical applicability in recommendation systems.

Contribution

It introduces a scalable parallel training approach for RBMs with Horovod, improving training efficiency and reducing time from hours to minutes.

Findings

Training time reduced to 12 minutes on 64 CPU nodes

Good scaling efficiency demonstrated

Enhanced practical usability of RBMs in recommendation systems

Abstract

Restricted Boltzmann Machine (RBM) is a generative stochastic neural network that can be applied to collaborative filtering technique used by recommendation systems. Prediction accuracy of the RBM model is usually better than that of other models for recommendation systems. However, training the RBM model involves Markov-Chain Monte Carlo (MCMC) method, which is computationally expensive. In this paper, we have successfully applied distributed parallel training using Horovod framework to improve the training time of the RBM model. Our tests show that the distributed training approach of the RBM model has a good scaling efficiency. We also show that this approach effectively reduces the training time to little over 12 minutes on 64 CPU nodes compared to 5 hours on a single CPU node. This will make RBM models more practically applicable in recommendation systems.