A COMPARISON OF CLASSIFICATION IN ARTIFICIAL-INTELLIGENCE, INDUCTION VERSUS A SELF-ORGANIZING NEURAL NETWORKS

Three methods of classification (machine learning) were used to produc e a program to choose a detector for ion chromatography (IC). The sele cted classification systems were: C4.5, an induction method based on a n information theory algorithm; INDUCT, which is based on a probabilit y algorithm and a self-organising neural network developed specificall y for this application. They differ both in the learning strategy empl oyed to structure the knowledge, and the representation of knowledge a cquired by the system, i.e., rules, decision trees and a neural networ k. A database of almost 4000 cases, that covered most IC experiments r eported in the chemical literature in the period 1979 to 1989, compris ed the basis for the development of the system. Generally, all three a lgorithms performed very well for this application. They managed to in duce rules, or produce a network that had about a 70% success rate for the prediction of detectors reported in the publication and over 90% success for choosing a detector that could be used for the described m ethod. This was considered acceptable due to the nature of the problem domain and that of the training set. Each method effectively handled the very high noise levels in the training set and was able to select the relevant attributes.