Reduced Training for Hierarchical Incremental Class Learning

Abstract

Hierarchical incremental class learning (HICL), proposed by Guan and Li in 2002, is a recently proposed task decomposition method that addresses the pattern classification problem. HICL is proven to be a good classifier but closer examination reveals areas for potential improvement. This paper presents an approach to improve the classification accuracy of HICL by applying the concept of reduced pattern training (RPT). The procedure for RPT is described and compared with the original training procedure. RPT systematically reduces the size of the training data set based on the order of sub-networks built. The results from benchmark classification problems show much promise for the improved model