Methods of combining multiple classifiers and applications to totally unconstrained handwritten numbers
Jia Li · Jisuanji gongcheng yu sheji · 2007
The problem of combining the outputs of several classifiers is encountered in various applications of pattern recognition.A new combination method of multiple classifiers is presented and the method is controlled by a feed-forward neural net trained with the backpropagation technique.When an unlabeled pattern is input to each individual classifier,it also goes to the neural net to decide two classifiers as the champion and the runner-up.And then let the two classifiers go to a randomizer to decide the final winner.The method is used to recognize totally unconstrained handwritten numerals.The experimental results show that the performance of individual clas- sifiers is improved considerably.