Improved inductive learning using training data reorganisation

This paper presents a solution to reorganize the training data set during learning process. Inductive learning from examples has been proposed as a measure to acquire knowledge automatically from expert systems. Inserting the example of each different class in the sequence of training examples carries out this reorganization process. The new method overcomes the problem of generating one default rule from the initial examples in the training data set. The results obtained after applying the reorganization method are superior to those produced by the original RULES-4 algorithm.