next up previous
Next: 3 Results Up: 2.2 Hand crafted fuzzy Previous: 2.2.1 Human details

2.2.2 Fuzzy logic toolkit

The fuzzy logic toolkit selected for the project was [Hillegas, Fuzzy Java]. This toolkit was written in a good language (Java), with reasonable documentation and good primitives for building fuzzy rules. Fuzzy rules in this toolkit took the form of ``tune'' statements, that are akin to a fuzzy version of the ``switch'' or ``case'' statements of traditional programming languages. In these statements, all the cases in the tune statement are executed to some degree, this degree is based on how true their condition is and their output is weighted accordingly.

A supporting program was written in Java to read in a dataset, apply the hand constructed, hard coded fuzzy rules to each instance, and then apply the defuzzification procedure to obtain the classification for that instance. The hard coding of the fuzzy rules proved adequate, but limited some possibilities; this is discussed in more detail in Section 4.2.2.

Kevin Pulo