Page 150 - Robots Androids and Animatrons : 12 Incredible Projects You Can Build

Short Medium Tall Medium
Short Tall

Gaussian Binary

Medium Medium

Short Tall Short Tall
129
Digitized Fuzzy
6.28 Grouping people by height using different schemes

one more example. If we graph the height of 1000 people, our
graph will resemble the first graph shown in Fig. 6.28. We can use
this graph of heights to classify the heights into groups: short,
medium, and tall. If we applied a hard rule that stated everyone un-
der 5′ 7″ is short and everyone taller then 6′ 0″ is tall, our graph will
resemble the second graph. This classifies a person who is 5′ 11.5″
tall as “medium,” when in actuality the person’s height is closer to
the tall (6′ 0″ and over) group.
Instead of hard-and-fast rules typically employed by computers,
people typically use soft and imprecise logic, or fuzzy logic. To
implement fuzzy logic in computers we define groups and quantify
the membership in that group. Groups overlap, as seen in the
fourth graph of Fig. 6.28. So the person who is 5′ 11.5″ tall is almost
out of the medium group (small membership) and well into the tall
group (large membership).
Fuzzy logic provides an alternative to the digitized graph shown as
the third graph of Fig. 6.28. A high-resolution digitized graph is
Team LRN Intelligence