Ameliorated Prediction of Breast Cancer using Hybridized Particle Swarm Optimization Techniques

Abstract

Breast cancer is a malignant tumor that develops, when cells in the breast tissue divide newlineand grow without the normal control on cell death and cell division. It is the most newlinecommon cancer among women. Breast cancer poses a serious threat and is the second newlineleading cause of death in women today. newlineBreast cancer prediction envisages whether it is present in patients and breast cancer newlinediagnosis distinguishes benign from malignant breast lumps. This research proposal newlinemainly concentrates on the diagnosis of breast cancer in women using Swarm Optimization newline(SO) since optimization is a process of reducing the task of misclassification newlinein prediction. Swarm Optimization has two types, namely Particle Swarm Optimization newline(PSO) and Ant Colony Optimization (ACO). Both the optimization techniques are newlineapplied in this research work. newlineWith SO as the basic technique, this research work portrays four different methodologies newlinewhich integrate Evolutionary Algorithm , Neural Networks , Classification newlineAlgorithm and Fire Ant Optimization (FAO) algorithm. The proposed methods have newlinebeen tested and validated withWisconsin Breast Cancer datasets from the UCI Machine newlineLearning Repository. The brief notes on the four different methods are as follows: newlineThe first method namely Genetic Algorithm (GA) is embedded in Particle Swarm newlineOptimization (PSO) - Hierarchical Radial Basis Function (HiRBF) and sorting, inserting newlineand removing techniques are predicted. The average accuracy of this method is newline98.4%. newlineThe second proposed model combines PSO with Non-Dominating Sorting (PSONDS), newlinemulti classifier techniques such as K-Nearest method, Fast Decision Tree and newlineKernel Density Estimation, and Bayesian Belief Network for revising the predicted newlineresults. This method yields an accuracy of 98.8%. newlineThe third method uses Fire Ant Optimization (FAO), which is a type of ACO. FAO newlinealong with Nearest Density Feature Selection (NDFA) increases the sensitivity, specificity, newlineaccuracy and reduces the time complexity of prediction. The average accuracy newlineobtained is 99.2%. newline