A Bayes Classifier Considering Environmental Change for Multivariate Signal Data

In this paper, we suggest learning algorithm of a high precision classifier for multivariate signal. The method deals with environmental influences. In this proposal technique, we define the features of the classification target and the environment as population parameters of probability distribution. We estimate the parameters by using the Bayesian inference. The Bayesian decision rule is used for the selection of similar environment properly in the proposed method. We try to evaluate the influence of the environmental change. In the numerical experiments, we verify that the proposed method has high classification accuracy. As the results, we show that our method can adapt environmental influence.