Generalized Oppositional Moth Flame Optimization with Crossover Strategy: An Approach for Medical Diagnosis

Abstract In the original Moth-Flame Optimization (MFO), search behavior of moth depends on corresponding flame and interaction between its flame, so it will get stuck in local optimum easily when facing multi-dimensional high-dimensional optimization problems. Therefore, this work, a generalized oppositional MFO with crossover strategy, named GCMFO, is presented to overcome mentioned defects. proposed GOBL employed increase population diversity expand range initialization iteration jump phase based rate; crisscross (CC) adopted promote exploitation and/or exploration ability MFO. The algorithm’s performance estimated by organizing series experiments; firstly, CEC2017 benchmark set evaluate GCMFO tackling multimodal Secondly, applied handle multilevel thresholding image segmentation At last, integrated into kernel extreme learning machine classifier deal three medical diagnosis cases, including appendicitis diagnosis, overweight statuses thyroid cancer diagnosis. Experimental results discussions show that approach outperforms other state-of-the-art algorithms both convergence speed accuracy. It also indicates has promising potential for application.