An application of feature selection to handwritten character recognition

The proposed feature selection method is applied to handwritten character recognition. This is based on an iterative factor selection method having two iterative loops. The outer loop is for selecting features necessary for error categories to be corrected. The inner loop is for selecting a subset suitable for the objective categories. This method introduces feature evaluation functions suitable for the objectives of each loop. It is experimentally applied to handwritten characters and is examined from several viewpoints