Hybrid Bayesian Network Structure Learning Using Map-Reduce For Big Data

Abstract

Predicting the future or outcome of any event has been in human’s nature since the beginning of time. We have always been trying to know what our actions will result in. With the advancement in the field of science and tech nology and with the increase in processing power of computer hardware, we have come very close to predicting certain outcomes based on prior knowl edge. Bayesian networks are one the ways of predicting an outcome. It falls into the category of Probabilistic Graphical Model. It finds it use in data mining and for representing uncertain knowledge. Big data, artificial intelli gence and machine learning rely on data and gets effected by changes in it. Bayesian Network helps in understanding the data and finding meaningful inferences, which are often basis of realistic applications. In this paper, we are going to discuss how Max-Min Hill Climbing, that is a hybrid algorithm, with Map-Reduce based framework can be implemented in order to lessen the execution time with similar accuracy