Automatic Case Generation for Pattern Classification

This paper shows how automatic cases can be generated using statistical learning methods such as support vector machines. This work also introduces new conflict resolution and revision strategies for enhancing the performance of the Case Based Reasoning system. The input to this system is a set of training samples out of which automatic cases are generated using support vector machines. These cases are subsequently used to classify unknown patterns and if a case is not found, it is added to the existing set of cases for future reference. In order to test the effectiveness of the proposed system, this algorithm was tested on a benchmark dataset and it was observed that it compares well with similar studies conducted.