Abstract
Artificial Neural Networks (ANNs) are models that learn patterns in input-output data. Since traditional optimization methods often get trapped in local optima when determining weight and bias values, identifying optimal parameters and enhancing network performance remain significant research areas. Heuristic algorithms are also generally used in solving optimization problems and are used to train ANNs. In the study, the parameter optimization of the ANN model was carried out using the Aquila Optimizer (AO), a recent metaheuristic algorithm, and a hybrid Aquila Optimizer optimized ANN model (AOANN) was proposed. Hybridization of algorithms contributes to the improvement of optimization performance. In this study, the proposed model was assessed on empirical datasets, including Cancer, Iris, Glass, and Wine, and its performance was compared with that of well-established ANN models. The results of the evaluation revealed that the proposed AOANN, a soft computation model, demonstrated stability in solving classification problems.