Falcon Optimization Algorithm for Bayesian Networks Structure Learning

Shahab Wahhab Kareem,Mehmet Cudi Okur

doi:10.7494/csci.2021.22.4.3773

Shahab Wahhab Kareem, Mehmet Cudi Okur

Open Access

https://doi.org/10.7494/csci.2021.22.4.3773

Copy DOI

Abstract

In machine-learning, one of the useful scientific models for producing the structure of knowledge is Bayesian network, which can draw probabilistic dependency relationships between variables. The score and search is a method used for learning the structure of a Bayesian network. The authors apply the Falcon Optimization Algorithm (FOA) as a new approach to learning the structure of Bayesian networks. This paper uses the Reversing, Deleting, Moving and Inserting operations to adopt the FOA for approaching the optimal solution of Bayesian network structure. Essentially, the falcon prey search strategy is used in the FOA algorithm. The result of the proposed technique is compared with Pigeon Inspired optimization, Greedy Search, and Simulated Annealing using the BDeu score function. The authors have also examined the performances of the confusion matrix of these techniques utilizing several benchmark data sets. As shown by the evaluations, the proposed method has more reliable performance than the other algorithms including producing better scores and accuracy values.

Highlights

Shahab Wahhab Kareem, Mehmet Cudi OkurOne of the simplified analytical models for constructing the probabilistic structure of knowledge in machine-learning is a Bayesian network (BN) [13]
The information-theoretic score is implemented in techniques such as the Akaike information criterion (AIC), log-likelihood (LL), minimum description length (MDL), Bayesian information criterion (BIC), mutual information test (MIT), and normalized minimum likelihood (NML) [3]
The authors concentrated on the structure learning of a Bayesian network problem and utilized the falcon-inspired optimization procedure for Bayesian network structure learning

Summary

Introduction

One of the simplified analytical models for constructing the probabilistic structure of knowledge in machine-learning is a Bayesian network (BN) [13]. There are various techniques of a research strategy that are intended to improve the problem of structural learning; these include particle swarm intelligence [4], the ant colony optimization algorithm [27], bee colony [13], the hybrid algorithm ( [11, 15, 21]), the simulated annealing algorithm [26], bacterial foraging optimization [33], genetic algorithms [19], the gene-pool optimal mixing evolutionary algorithm (GOMEA) [24], the breeding swarm algorithm [18], the binary encoding water cycle [32], pigeon-inspired optimization [16], tightening bounds [6], A* search algorithms [34], scatter search documents [5], the cuckoo optimization algorithm [1], quasi-determinism screening [25], and the minimum spanning tree algorithm [28] Another additional metaheuristic technique that can be applied to learn the structure of Bayesian networks is falcon optimization.

Bayesian network structure learning

Falcon optimization algorithm

Structure learning of Bayesian network using FOA

Experimental evaluation

Conclusion