Improving classification performance of cancer microarray data using hybridization of binary grey wolf and particle swarm optimization

In this study, we have proposed hybridization of binary grey wolf Optimization and particle swarm optimization (BGWOPSO) method and we compared this hybrid optimization method with Particle Swarm Optimization (PSO) and Binary Grey Wolf Optimization (BGWO). We have used five significantly different classifier such as K-nearest Neighbor (KNN), Support Vector Machine (SVM), Artificial Neural Network (ANN), Logistic Regression (LR), Random Forest (RF). Furthermore, we have used ratio comparison validation for the 10-folds cross-validation method for feature selection methods. Data sets such as Leukemia, Breast Cancer, and Liver Cancer are used to apply the combinations and measure accuracy as well as the area under ROC. Moreover, the results show that Hybrid optimization method (BWOPSO), significantly outperformed the both binary grey wolf optimization (BGWO) and particle swarm optimization (PSO) method, when using several performance measures including accuracy, selecting the best optimal features. Secondly, combinations of classifier and feature pre-processing method significantly improve the accuracy. Lastly, the AUC value is been displayed in this study.

Name thesis:degree_name

Master of Science (MSc) in Computational Sciences

Grantor dc:publisher

Laurentian University of Sudbury

Year dc:date.issued

2019

Author dc:creator

• Savaliya, Leena

• Hybrid binary optimization
• Grey wolf optimization
• Particle swarm optimization
• Feature
• classification.

Language dc:language.iso

en

Repository record dc:identifier.uri

https://laurentian.scholaris.ca/handle/10219/3584

OAI record oai:identifier

https://laurentian.scholaris.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Alaurentian.scholaris.ca%3A10219%2F3584