Incremental support vector machines for handwritten arabic character recognition
H. Bentounsi, M. Bataouche · 2004
In this paper, the use of support vector machines (SVM) for handwritten Arabic character recognition is studied. SVMs are based on structural risk minimization, which tries to maximize the generalization capability on the unseen data by reducing the empirical risk on the seen data. SVMs are able to summarize the data space in a very concise manner as support vectors, for incremental learning by preserving at each step of training the resulting support vectors is used. The model obtained by this method is the same or similar that has been obtained using all the data.