A Novel Evolutionary Algorithm for Data Classification Problem With Extreme Learning Machines

Machine learning techniques have gained great popularity due to their success in data classification problems. This study proposes a novel evolutionary feature selection algorithm integrated with Single Hidden Layer Feed-forward Neural Networks (SLFN)s. Our main goal is to find out the most efficient subset of features and provide the best prediction accuracy. The algorithm combines the evolutionary technique of genetic algorithms (GA) and calculates the fitness values (prediction accuracy) of each selected subset of features by using Extreme Learning Machines (ELM). The results of the SLFN are calculated in a faster manner, which is very suitable for the GA while optimizing the selection of the best subset of features. The experimental results show that the proposed algorithm provides significant improvements. Competitive results are obtained/verified by comparing our solutions with those of the state-of-the-art data classification algorithms.