Hybrid Gradient Descent Grey Wolf Optimizer for Cloud Workload Balancing with Optimal Feature Selection using Reinforcement Learning

Abstract

Method of reducing the elements from a dataset by removing irrelevant, redundant, and randomly selected features which is called feature selection. It aims to reduce training time and improve data quality especially for big and complex datasets. This study introduces an optimizer for feature selection problems by combining the metaheuristic algorithm called the grey wolf optimizer with gradient descent algorithm. The proposed approach outperformed the original grey wolf optimizer on various test functions and showed promising results on clinical datasets from the UCI machine-learning repository. It suggests potential by enhancing feature selection techniques in data analysis.