Due to the data explosion, Big Data is everywhere all around of us. The curse of dimensionality
in Big Data has produced a great challenge for data classification problems. Feature
selection is a crucial process to select the most important features to increase the
classification accuracy and to reduce the time complexity. Traditional feature selection
approaches suffer from various limitations, so Particle Swarm Optimization (PSO)-based
feature selection approaches are proposed to overcome these limitations, but classical
PSO shows premature convergence when the number of features increases or the datasets
having more categories/classes. In this paper, topology-controlled Scale-Free Particle
Swarm Optimization (SF-PSO) is proposed for feature selection in high-dimensional
datasets. Multi-Class Support Vector Machine (MC-SVM) is used as a machine learning
classifier and obtained results show the superiority of our proposed approach in big
data classification.
