Learning algorithms with boosting for vector quantization

There have been proposed many learning algorithms for VQ based on the steepest descend method. However, any learning algorithm known as a superior one does not always work well. This paper proposes a new learning algorithm with boosting. Boosting is a general method which attempts to boost the accuracy of any given learning algorithm. The proposed method consists of three sub-learners. The first sub-learner is constructed by performing the conventional learning algorithm with data randomly selected from given data space. The second sub-learner is constructed by performing the conventional learning algorithm with data selected with higher probability from data incorrectly learned by the first sub-learner. The third sub-learner is constructed with data for which either the first or the second sub-learner is incorrectly learned. That is, the method attempts to construct different kinds of reference vectors by using different kinds of data sets constructed from the original data set. The output for any input data is given as decision by averaging the outputs of three sub-learners. In order to show the effectiveness of the proposed algorithm, numerical simulations are performed.