Particle Swarm Optimization for classification of breast cancer data using single and multisurface methods of data separation

This paper exploits the simplicity, efficiency and flexibility of the particle swarm optimization (PSO) method to propose a single and multisurface based data separation methods for classification of Breast Cancer Data. Like most artificial intelligence based techniques the first step of the proposed approaches involve the training of the PSO-based classifiers according to pre-defined data separation methods, using part of the dataset for training. The performances of the classifiers are then tested on the remaining dataset to measure the classification accuracy. The training and testing datasets are derived from the Breast Cancer database obtained from the UCI machine learning repository. Both separation methods produce good classification performance; however, the method based on multiple separating surfaces achieves the best result of 100% classification accuracy on both the training and testing datasets.