A memory based classifier using the recursive partition averaging

Proposes the RPA (Recursive Partition Averaging) algorithm in order to improve the storage requirements and classification time of the memory-based reasoning method. The proposed method enables us to use the storage more efficiently by extracting representatives from training patterns. After partitioning the pattern space recursively, it averages patterns in each hyper-rectangle to extract a representative. Also, we have used the mutual information between the features and classes as weights for the features, in order to improve the classification performance. Experimental results show that RPA is superior to K-NN (K-nearest neighbors) and the EACH system in terms of memory usage and classification accuracy