Toward One Class Classifier techniques applied to verifier information

One Class Classifier techniques have the ability to identify data that is unknown with respect to a group of known observations. However, the training set only contains instances of the known class and no instances at all or very few instances of unknown data. During training, in the verify problem of a wrapper, we only have instances of the classes we know. Therefore, the One Class Classifier techniques could be applied. In order to evaluate the performance of these methods we use different databases proposed in the current literature. Statistical analyses of the results obtained by some basic One Class Classification techniques like parzen_dd, gauss_dd, svmdd, som_dd and knndd will be described.