Fuzzy Logic vs Machine Learning
An Investigation

Kevin Pulo, login kpulo
kev@zip.com.au
http://www.zip.com.au/ kev/

June 1999

Abstract:

Fuzzy logic has allowed subjective, real-world imprecisions to be modelled with high degrees of accuracy, compactness and ease. However the fuzzy rules (most often used in control systems) have, for the most part, been constructed manually by experts in the problem domain.

This project aimed to investigate the effectiveness of standard machine learning algorithms when compared to human based learning processes, expressed in concrete terms by fuzzy rules. The results highlight several advantages standard machine learning algorithms have over humans, and suggest numerous directions in which fuzzy advances could be made to improve the algorithms.