A NOTE ON COMPARING CLASSIFIERS

Recently many new classifiers have been proposed, mainly based on neur al network techniques. Comparisons are needed to evaluate the performa nce of the new methods. It is argued that a straightforward fair compa rison demands automatic classifiers with no user interaction. As this conflicts with one of the main characteristics of neural networks, the ir flexibility, the question whether they are better or worse than tra ditional techniques might be undecidable.