An Efficient Hybrid Genetic-Grey Wolf Based Neural Network (G2NN) for Breast Cancer Data Classification

Machine learning is the one of the famous Artificial Intelligence (AI) technique. Data Mining or Machine Learning techniques are most popular in medical diagnosis, classification, forecasting etc. K-Nearest Neighbor, SVM (Support Vector Machine), DT (Decision Tree),RF (Random Forest),NN (Neural Network) are famous classification algorithms. Neural Network is one of the popular techniques, which is used to refine the verdict of breast cancer. A neural network is otherwise known as Artificial Neural Network(ANN), which is mimicking of biological neurons of human brain. Genetic Algorithm (GA) is emerged bio inspired technique. Selection, Crossover, and Mutation are three operations in Genetic Algorithm. The performance of a genetic algorithm depends on the genetic operators, particularly crossover operator. Grey Wolfoptimization algorithm is inspired from hunting of wolf strategy. Alphas, Beta, Gamma are the three levels of processes. In this paper, a novel hybrid Genetic Grey Wolf based Neural Network is introduced and we named it as G2NN. In the field of medical, we need more accuracy when compared to other field, because it relates to human life. Many researchers found new novel ideas for breast cancer data classification using neural network model. Among many diseases, Breast Cancer is one of the unsafe diseases among women in India and in addition to the whole world. The early detection of cancer helps in curing the disease completely. In many research areas Genetic Algorithm and Grey wolf algorithm are used to train neurons in order to yield good accuracy. In this manuscript, a new Genetic Grey Wolf optimizer based Neural Network is introduced and we compare the proposed work with other techniques like SVM(Support Vector Machine),NN (Neural Network), Genetic based Neural Network, Grey wolf based Neural Network and the experimental results of proposed work produced better result. The proposed algorithm produces 98.9 % of accuracy on UCI Wisconsin breast cancer dataset.