Category Trees - Classifiers that Branch on Category

This paper presents a batch classifier that splits dataset into tree branches depending on the category type. It has been improved from earlier version and fixed mistake in paper. Two important changes have made. The first is to represent each with separate classifier. Each then classifies its own subset of data rows, using input values create centroid also itself. If contains more than one however, it needs new classifiers for incorrect data. second change therefore allow branch layers when there split data, rows are incorrectly classified. layer can like - not distinguishing features, but categories. suggests further innovation, which some columns value ranges, or bands. When considering shown be classified directly through while rest must technique idea allows discuss biological analogy neurons neuron links. Tests show method successfully classify diverse set benchmark datasets better state-of-the-art.