Binary Duck Travel Optimization Algorithm for Feature Selection in Breast Cancer Dataset Problem

Abstract

To predict if a tumor is malignant or benign is the challenge to all the researchers achieving efficient prediction of women in breast tumor. Unrestrained cell expansion in hankies of the breast is called breast cancer disease. Successful detection of breast cancer in earlier stage is required to secure women against high mortality. To select the best features among breast cancer classification of benign/malignant from the input mammogram images originated from Wisconsin dataset is satisfying with the new algorithm binary duck travel optimization algorithm (bDTO). Sigmoid activation function (logistic regression) is used for binary classification of proposed method. Food forage activation of ducks position is updated by sigmoid with maximum likelihood function (SMLE) of entire duck flock. Classifying the given mammogram is cancer or non-cancerous is based on the optimal feature selection by bDTO through (SMLE). Feature extraction of mammogram Corpus is done with the aid of sigmoid activation function (SAF) classifier. (bDTO-SMLE-SAF) is an intrinsic procedure to eliminate irrelevant scope and select the optimal highlights by using Wisconsin families normal nucleoli, single epithelial cell size, bare nuclei, uniformity of cell size, uniformity of cell shape, bland chromatin, mitoses, marginal adhesion, clump thickness features that are evaluated by the quality measures exactness, compassion, specificity, accuracy, evoke, and F-value clearly shows that bDTO classifier has the maximum accuracy 92% while compared with the DTO and SAF classifiers. The efficiency of an algorithm is proved by the promising results for selecting the best feature of malignancy classification through bDTO algorithm.