Fast Adaptive Classifier based on a Stepwise algorithm

In order to design a classifier that is not only fast and accurate, but also adaptive, the fast adaptive classifier (FAC) based on the model of Adaptive Hinging Hyperplanes (AHH) is proposed. The algorithm for training FAC is developed based on the AHH algorithm, and the training speed is greatly improved. Results of numerical experiments are reported, and these results demonstrate that FAC can achieve good classification accuracy on most data sets. In addition, FAC enjoys a greater advantage that it combines feature selection and classification together, i.e., feature selection is also fulfilled in the classification process. Moreover, thanks the stepwisely training strategy, FAC also enjoys the unique property of adaptiveness, which results in a simple model structure.