Algorithmic framework based on the binarization approach for supervised and semi-supervised multiclass problems

Abstract

Using a set of binary classi ers to solve the multiclass classi cation problem has been a popular approach over the years. This technique is known as binarization. The decision boundary that these binary classi ers (also called base classi ers) have to learn is much simpler than the decision boundary of a multiclass classi er. But binarization gives rise to a new problem called the class imbalance problem. Class imbalance problem occurs when the data set used for training has relatively less data items for one class than for another class. This problem becomes more severe if the original data set itself was imbalanced. Furthermore, binarization has only been implemented in the domain of supervised classi cation. In this thesis, we propose a framework called Binarization with Boosting and Oversampling (BBO). Our framework can handle the class imbalance problem arising from binarization. As the name of the framework suggests, this is achieved through a combination of boosting and oversampling. BBO framework can be used with any supervised classi cation algorithm. Moreover, unlike any other binarization approaches used earlier, we apply our framework with semi-supervised classi cation as well. BBO framework has been rigorously tested with a number of benchmark data sets from UCI machine learning repository. The experimental results show that using the BBO framework achieves a higher accuracy than the traditional binarization approach.