Opposition-based Harris Hawks optimization algorithm for feature selection in breast mass classification

One of the very common forms of cancer found in women all over the world these days is Breast cancer. Mammography is regarded as the general way to detect this form of cancer at an earlier stage. The radiologist uses the mammogram images to search for the abnormalities present in the breasts via visual perception only. Masses are considered as one of the most common abnormalities, which are a collection of cells bounded together in a more dense way than the tissues surrounding it, which may be of benign or malignant nature. These days Machine learning-based Computer-aided diagnosis Systems assist the radiologist to make the right prediction, however there still lays a scope of improvement in the techniques for selecting the optimal number of features with maximum accuracy. Selecting the optimal size of feature subset from a larger number of features is considered as an optimization problem. This research proposes the Opposition based variant of one of the most recent optimization algorithm known as Harris Hawks Optimization algorithm. Results indicate that the proposed algorithm performs better than other algorithms on various benchmark functions and on various criteria's considered for feature selection in breast mass classification.