A study on rough null space based support vector machine

摘要

Automatic document classification is becoming an important research field with the rapid increase of electronic documents. The main purpose of this research is to construct an accurate document classifier based on support vector machines (SVM), which is known as the state of the art algorithm for document classification. The rough null space (RNS) based approach is also known as a good linear approach for image recognition. The question is, can we combine RNS with SVM, and obtain a better system for document classification? In this paper, we introduce the basic idea of RNS+SVM, and compare it with SVM using experimental results.