Harris Hawks Optimization Based Feature Selection on High Dimensional data
Suvendu Kumar Jena, Gotte Ranjith Kumar, Ramy Riad Al–Fatlawy, S. Rasheed Mansoor Ali, T M Aruna · 2024
In recent times, evolutionary algorithms have been demonstrating significant benefits in feature selection due to their simplicity and global search capability. However, most of the existing evolutionary algorithms are computationally intensive for high-dimensional data. In order to decrease computational costs, effective evaluation methods are crucial. In this paper, Harris Hawks Optimization (HHO) technique is proposed for effective feature selection on high dimensional data. Firstly, a Multi-Layer Perceptron (MLP) is used to form an integer optimization problem for efficient feature selection process. Then the proposed HHO method is applied for the process of feature selection. This method has three steps such as Initialization, solutions update and classification. The performance of proposed HHO algorithm is compared with the existing classifiers such as Binary Waterwheel Plant Algorithm (BWWP A) and Community detection based Genetic Algorithm (CDGAFS) and achieved 98.2% of accuracy.