Support Vector Machine Classification Method Based On Rough Set

The theory of support vector machine (SVM) is based on statistical learning theory, is a relatively new machine learning theory. It better solves the nonlinearity, high dimension and local minima problems, now become a new hotspot. Incremental SVM training algorithm can solve the massive sample problem that traditional support vector machine training algorithm can not solve, and can revise existing training support vector machines to avoid the waste of resource caused by re-learing. Rough set theory is a mathematical tool to deal with incomplete and uncertain knowledge, it can reduce knowledge under maintained the same classification ability of decision table. This paper combines rough set theory and incremental support vector machine training algorithm, using rough set theory for data preprocessing, researches the based on rough set incremental classification method of support vector machine.The genetic algorithm is an efficient optimization tool, The adaptive operator of adaptive genetic algorithm can adjust the crossover probability and mutation probability, to speed up the convergence. In this paper, the improved adaptive genetic algorithm is used for the attribute reduction, an attribute reduction algorithm based on improved adaptive genetic algorithm is proposed.In Incremental SVM training process, how to eliminate useless samples and retaining useful samples was a key problem. In this paper, the distance ratio method is used for sample selection, and k nearest neighbor method is used to extract boundary sample. This paper presents an incremental support vector machine training algorithm, the algorithm based on the distance ratio method and the k nearest neighbor method. To combine the attribute reduction algorithm and the incremental training algorithm of support vector machine, a support vector machine classifier based on rough set is constructed. The effectiveness and performance of the presented algorithms are verified by experiment and comparative analysis.