Fuzzy Future, From Society and Science to Heaven in a Chip, Bart Kosko, New York, 1999, Harmony Books, a member of the Crown Publishing Group.

A very philosophic look at fuzzy logic principles and the world in general.  On the actual programming end of it, there are three ideas involved in, or rather, with fuzzy logic: neural networks, rules and fuzzy logic itself.

Fuzzy logic refers to, rather than saying a thing is black or white, acknowledging that it is somewhat black and somewhat white, that a test is somewhat true and somewhat false, to concepts like warm, bright, early.  Fuzzy logic tests to what degree a thing is true and what degree it is false.

The rules tell the logic how to interpret the evaluation of its tests.  If a thing is more white than black and it is before 12:00 then do this.  As the inputs increase (temperature, time, brightness, etc.) the number of rules tend to explode.

Rather than relying on an expert to enter the rules, neural systems learn by themselves through pattern recognition.  The trouble here, is that if a wrong thing happens, the learned rules are all skewed.

© Lester L. Noll

21-Apr-2001