Quasi Opposition–Based Learning With Artificial Rabbit Optimization for Feature Selection of Breast Cancer

Mammogram images are the most appropriate modality for breast cancer (BC) diagnosis which supports early diagnosis in women. However, the handcrafted features are insufficient for effective BC classification, which negatively impacts the model's accuracy and leads to the adaptive optimization problem. To overcome this problem, this research proposes Quasi Opposition–based Learning with an Artificial Rabbit Optimization (QOBL‐ARO) approach for the feature selection process in BC classification. Then, a Grid Search–based Support Vector Machine (GS‐SVM) approach is introduced for the classification of BC into binary classes. Through utilizing GS to optimize the kernel selection, GS‐SVM captures complex patterns in the data, making it highly effective at distinguishing between benign and malignant BCs. A significance of the QOBL‐ARO approach with the GS‐SVM approach is executed by using two standard datasets such as MIAS and DDSM. The experimental findings specify that the proposed QOBL‐ARO approach with the GS‐SVM approach reaches an accuracy of 99.45%, 98.55%, and 99.69% on MIAS, DDSM, and INBreast datasets individually as compared to existing SVM and Extreme Learning Machine with Crow Search Algorithm, namely (ICS‐ELM).