Improving Golden Jackal Optimization Algorithm Based on Elite Opposition-Based Learning: A Case Study on Breast Cancer Histopathology Images Classification

Abstract

Feature extraction and categorization of histopathological images are crucial for predicting and diagnosing diseases such as breast cancer. A common problem with the characteristic matrix is that it includes information that is unrelated to the context. The selection of relevant characteristics has been shown to improve the efficacy of multiple categorization methods substantially. The Metaheuristic algorithm “Golden Jackal Optimization” (GJO) effectively emulates the cooperative hunting strategy of jackals. Nonetheless, it is susceptible to converging on a local optimum, since the prey’s position update mostly depends on the male golden jackal, and there is a lack of diversity among golden jackals in certain situations. The Elite Opposition-Based Learning (EOBL) technique enhances the distribution of initiated solutions throughout the search space. Furthermore, the use of the EOBL method improves the algorithm’s computational precision and expedites its convergence rate. The objective of this study is to enhance the efficacy of the GJO algorithm by utilizing the EOBL to improve the classification accuracy of histological images of breast cancer. Experimental results using publicly available breast cancer histopathology image datasets show that the proposed approach substantially surpasses competing algorithms in terms of overall Accuracy, Precision, Recall, and F1-score metrics. The implemented approach proved beneficial for medical picture categorization in actual clinical settings.