An Efficient Search Strategy for Aggregation and Discretization of Attributes of Bayesian Networks Using Minimum Description Length

TL;DR

This paper introduces an efficient search strategy for discretizing attributes in Bayesian networks, enabling the practical application of a minimum description length-based discretization method that preserves data information.

Contribution

It presents a novel, highly efficient search algorithm that makes the MDL-based discretization approach feasible for larger Bayesian networks.

Findings

The proposed method significantly reduces computational complexity.

It effectively preserves data information during discretization.

The approach is applicable to large-scale Bayesian network structure recovery.

Abstract

Bayesian networks are convenient graphical expressions for high dimensional probability distributions representing complex relationships between a large number of random variables. They have been employed extensively in areas such as bioinformatics, artificial intelligence, diagnosis, and risk management. The recovery of the structure of a network from data is of prime importance for the purposes of modeling, analysis, and prediction. Most recovery algorithms in the literature assume either discrete of continuous but Gaussian data. For general continuous data, discretization is usually employed but often destroys the very structure one is out to recover. Friedman and Goldszmidt suggest an approach based on the minimum description length principle that chooses a discretization which preserves the information in the original data set, however it is one which is difficult, if not impossible, to implement for even moderately sized networks. In this paper we provide an extremely efficient search strategy which allows one to use the Friedman and Goldszmidt discretization in practice.