In many text classification applications, it is appealing to take every document as a string of characters rather than a bag of words. Previous research studies in this area mostl...
A binary hierarchical classifier is proposed to solve the multi-class classification problem. We also require rejection of non-target inputs, which thus producing a very difficult...
Abstract. Standard Support Vector Machines (SVM) text classification relies on bag-of-words kernel to express the similarity between documents. We show that a document lattice can ...
In this paper, a high-speed document image classification algorithm is presented. The algorithm is based on the bottom-up strategy which can successfully segment and classify any ...
It is difficult to apply machine learning to new domains because often we lack labeled problem instances. In this paper, we provide a solution to this problem that leverages domai...