Multi objective Cost sensitive subspace reduction technique for access control system

A Cost sensitive subspace reduction technique using multi objective optimizer has been designed for an access control system having a hierarchy of classes starting from the most important class to the least desirable class. The main aim of this paper is to extract the cost sensitive features from facial images of test subjects and project the faces on to a domain with reduced dimension. The feature vector consisting of the cost sensitive information has been obtained by assigning appropriate cost values for different classes so that the occurrence of false rejection is higher than that of false acceptance. The reduced feature vectors were trained using neural network classifier. The proposed method has been found to be better than many contemporary methods. The percentage value of false acceptance obtained is of the order of 1 to 2 which is very less so far as suppressing the acceptance of less desirable classes is concerned.