Incremental Learning with Support Vector Machines and Fuzzy Set Theory

Over the past few years, a considerable number of studies have been made on Support Vector Machines (SVMs) in many domains to improve classification or prediction. However, SVMs request high computational time and memory when the datasets are large. Although incremental learning techniques are viewed as one possible solution developed to reduce the computation complexity of the scalability problem, few studies have considered that some examples close to the decision hyperplane other than support vectors (SVs) might contribute to the training process. Consequently, we propose a novel algorithm by improving Syed’s incremental learning method based on fuzzy set theory. At each learning step, SVs and potential informative examples, called candidate examples (CEs), are added to the next incremental learning step. We expect to achieve better accuracy and less execution time than other methods. In this ongoing study, the proposed algorithm would be investigated on five standard machine learning benchmark datasets to demonstrate the effectiveness of the method.