Case adaptation using estimators of neural network

One important issue in case-based reasoning system is the adaptation of cases. We proposed a new method for case adaptation, which is based on the estimators of ANN. Using m similar cases to the query case, a radial basis function ANN (RBF network) is trained to provide solutions for the query case. And then an interval solution with certain confidence is obtained by combining the estimators of the ANN model and the proposed solution provided by the RBF network. Experiment shows this approach greatly improves the accuracy of case adaptation by considering the estimators of ANN model.