Probabilistic Classifier and Its Application to Face Recognition

This paper proposes a method to classify different subjects from a large set of subjects. Taking correct decision in the process of classification of various subjects from the large set is an arduous task, since its probability is very low. This task is made simple by the proposed Probabilistic Classifier (PC). Maximum Likelihood Estimation (MLE) and Error Minimizing Algorithms (EMA) are the basis for the proposed classifier. Interpreting the EMA output in a probabilistic manner gives rise to PC. Concept of feedback is used in the classification process to enhance the decision rule. Experimental results obtained by applying the proposed classifier on various benchmark facial datasets, show its promising performance. Eventually, PC is found to be independent of the datasets.