DEVELOPMENT OF HYBRID EVOLUTIONARY ALGORITHMS

Abstract

Thirdly, PSO and the Chameleon Search Algorithm (CSA) are combined to create a hybrid algorithm called the Chameleon Search Algorithm-Particle Swarm Optimiza tion (CSAPSO). To improve optimization efficiency, the CSAPSO algorithm integrates Opposition-Based Learning with Hybrid. The proposed CSAPSO algorithm has been evaluated on twelve chest X-ray (CXR) and COVID-19 CXR images. CSAPSO’s performance is assessed by contrasting its outcomes with other state-of-the-art opti mization algorithms and other deep learning models using measures like Root Mean Square Error (RMSE), Peak signal-to-noise ratio (PSNR), and Structure Similarity In dex (SSIM), Classification Accuracy, Area Under Curve (AUC). Finally, a hybrid optimization algorithm called the Opposition-based Particle Swarm Algorithm-Grey Wolf Optimization algorithm (Opp-PSOGWO) is developed by inte grating Figurate Opposition-Based Learning (OBL) into the hybridization of PSO and Grey Wolf Optimizer (GWO). Opp-PSOGWO generates solutions in opposite direc tions, in the Fibonacci sequence, leading to an optimal initial population with increased diversity. The resultant hybrid algorithm can escape local optima traps, and Figurate Opposition-based Learning speeds up the search for the best solutions. Nine traditional benchmark functions (unimodal and multimodal) were used to test the Opp-PSOGWO model. Then, it has been compared to its parent algorithms, PSO and GWO, taking into account performance metrics, including mean, standard deviation, and CPU time. In conclusion, this research provides four unique algorithms that contribute to swarm based EA techniques. Compared to traditional algorithms, the hybrid BFA-FA, parameter free PSOBOA, hybrid CSAPSO, and Opp-PSOGWO algorithms enhance optimization performance in various real-world applications.