Successive Pattern Learning Based on Test Feature Classifier and Its Application to Dynamic Recognition Problems

Fundamental characteristics of successive lerarning by Test Feature Classifier(TFC), which is non-parametric and effective with small data, are examined. In the learning, a new set of training objects, they are fed into the classifier in order to obtain a modified classifier. We propose an efficient algorithm for reconstruction of prime test features, which are combinaton feature subsets for getting the excellent performance. We apply the proposed successive TFC to dynamic recognition problems where the traning data increase successively and also characteristic of the data change with progress of time, and examine the characteristic by the experiments which used the real world data and a set of simulated data.