LAUSR

Models based on machine learning algorithms have been developed to detect the breast cancer disease early. Feature selection is commonly applied to improve the performance of these models through selecting only relevant features. However, selecting relevant features in unsupervised learning is much difficult. This is due to the absence of class labels that guide the search for relevant information. This kind of the problem has rarely been studied in the literature. This paper presents a hybrid intelligence model that uses the cluster analysis algorithms with bio-inspired algorithms as feature selection for analyzing clinical breast cancer data. A binary version of both moth flame optimization and whale optimization algorithm is proposed. Two evaluation criteria are adopted to evaluate the proposed algorithms: clustering-based measurements and statistics-based measurements. The experimental results positively demonstrate that the capability of the proposed bio-inspired feature selection algorithms to produce both meaningful data partitions and significant feature subsets.