A framework for selecting salient features and samples simultaneously to enhance classifier performance

It is desirable to select out the salient subset of features and remove from the training set the instances that are not helpful to forming the final decision function of classifier. In present work we are trying to increase the classifier performance through efficiently selecting features and samples simultaneously. A new framework that coordinates feature selection and sample selection together is built. The criteria of optimal feature selection and the method of sample selection are designed. Using benchmark datasets, the effectiveness of the framework was tested in terms of their ability to raise the classifying correct rate while reducing the size of attribute set. Experimental results show that this new framework is effective and practical.