FUZZY CONTROL

1
???????? FUZZY
2
????????
3
????????
???????????????????????????????????? Rule 1 If
feature1 is high and feature2 is low and feature3
is medium, then class is 1 with confident CF1
Rule 2 If feature1 is medium and feature2 is
low and feature3 is then class is 3 with
confident CF2 Rule 3 If feature1 is high and
feature2 is high and feature3 is medium, then
class is 2 with confident CF3.
4
????????
Simulated 3-Class 2-Dimension Problem This
simulated problem is a linearly separable
3-classe 2-dimension pattern space. There are 80
patterns all together generated. Class 1 has 20
patterns class 2 has 20 patterns and class 3
has 40 patterns. The plot of this data set is
shown below.
5
????????
For a 3-class, 2-dimension problem above, if we
use 2 fuzzy linguistic label, small and large, we
can construct a 4, i.e., 22, fuzzy rules for a
fuzzy expert system as Rule 1 If x1 is small
and x2 is small then class is 2 (). Rule 2 If
x1 is small and x2 is large then class is 1 (?).
Rule 3 If x1 is large and x2 is small then
class is 3 (?). Rule 4 If x1 is large and x2 is
large then class is 3 (?). This can be
automatically generated by using the ILFN
classifier and then use ilfn2rule to map the
ILFN weight to fuzzy if-then rules.