Performance Assessment of Meta-Heuristic and WOA Algorithms for End-to-End Connection Weights Enhancement of (DeepLABNet)

Radial Basis Function (RBF) neural networks are widely used in artificial intelligence applications. However, traditional training algorithms often suffer from local optima stagnation and slow convergence, reducing network efficiency. This study investigates using the Whale Optimization Algorithm (WOA) to optimize RBF neural networks for improved cancer classification accuracy. WOA addresses the limitations of traditional algorithms by avoiding local optima and ensuring global nonlinear parameter optima. We evaluated WOA-RBF on cancer datasets and compared its performance with conventional algorithms, including Genetic Algorithm, Particle Swarm Optimization, Ant Colony Optimization, and Differential Evolution. Results indicate WOA-RBF outperforms other algorithms in precision and convergence, achieving 97% average classification accuracy for ovarian tumours and 96% for pancreatic tumours. WOA-RBF demonstrated faster convergence and better generalization capabilities. This study fills the gap in optimizing RBF neural networks using WOA, providing a reliable solution for cancer classification tasks. This research contributes to developing more accurate cancer diagnosis systems using RBF neural networks optimized with WOA.