Social Group Optimization-Based Neural Networks for Classification of Dataset

Evolutionary computation is a collection of algorithms based on the evolution of a population toward a solution to a certain problem. These algorithms have demonstrated their effectiveness in various optimization tasks. Notably, they have been extensively utilized to enhance the learning capabilities of classifiers, especially in the context of artificial neural network (ANN) Classifiers. Slow convergence and constant trapping at the local minima are two major issues with ANN classifiers. To address this issue, social group optimization (SGO) has been employed to identify the optimal parameters for the learning mechanism. In this work, a feedforward neural network (FNN) was fitted with SGO to improve learning and enable lower error rates and more precise categorization of high-dimensional datasets. Three distinct programs were developed: differential evolution with NN (DENN), particle swarm optimization with NN (PSONN), and teaching learning-based optimization with NN (TLBONN). These programs were employed to investigate the impact of these optimization methods on social group optimization with neural network (SGONN) learning across a variety of datasets. The results of this investigation revealed that SGONN has exhibited highly promising outcomes and manifested notably smaller errors when compared to the results achieved by PSONN, DENN, and TLBONN.