Logic
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Sub-topics:
Fuzzy sets
Operations with fuzzy sets
Geometrical representation of fuzzy sets
Fuzziness versus probability
Conceptualising in fuzzy terms
The extension principle
Fuzzy relations and fuzzy implications
Fuzzy propositions and fuzzy logic
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy Sets
Figure 3.1
Membership functions representing three fuzzy
sets for the variable "height".
Fuzzy Sets...
Figure 3.2
Representing
crisp and
fuzzy sets as
subsets of a
domain
(universe) U".
Fuzzy Sets...
Figure 3.3
Support of a
fuzzy set A
Fuzzy Sets...
Figure 3.4
I-cut of a
fuzzy set
Operations with fuzzy sets
Figure 3.5
Five operations with two fuzzy sets A and B
approximately represented in a graphical form
Operations with fuzzy sets...
Figure 3.6
Showing graphically one way to measuring similarity and distance
between fuzzy sets A and B. The black area represents
quantitatively the measure.
Operations with fuzzy sets...
Figure 3.7
A graphical
representation
of similarity S
(the black area)
between two
fuzzy sets B
and A based on
possibility P
and necessity N
measures (see
the formulas in
the text).
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Geometrical representation of
fuzzy sets
Figure 3.8
A geometrical
representation of the
concept of "heavy
smoker" from The
Smoker and the Risk
of Cancer case
example as a point in
a (three dimensional)
cube and as a (two
dimensional)
membership
function.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzziness versus probability
Figure 3.9
Probability
density
function for
throwing a
dice and the
membership
functions of
the concepts
"Small"
number,
"Medium",
"Big".
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Conceptualising in
fuzzy terms
Standard membership functions:
• single-valued, or singleton
• triangular
• trapezoidal
• S-function (sigmoid function):
– S(u) = 0, u<=a
– S(u) = 2((u-a)/(c-a))2 , a <u <= b
– S(u) = 1 - 2((u-a)/(c -a))2 , b <u <= c
– S(u) = 1, u > c.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Conceptualising in
fuzzy terms...
more standard membership functions...
• Z function:
– Z(u)= 1 - S(u)
• Pi - function:
• P(u)=S(u), u<=b; P(u)=Z(u), u>b.
Figure 3.12
One
representation
for the fuzzy
number "about
600".
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Conceptualising in fuzzy
terms...
Figure 3.13
Representing
truthfulness
(certainty) of
events as
fuzzy sets over
the [0,1]
domain.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
The extension principle
Figure 3.14
An illustration of
the extension
principle in fuzzy
theory: an
example of
transforming a
fuzzy set "x is
about 2" into a
fuzzy set "f is
about f(2)" for
f(x)= (x 1)2.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy relations and fuzzy
implications
Figure 3.15
Several useful fuzzy implications. The following short denotations
are used: u instead of n(u); v instead of n(v); "n" is minimum; "n"
is maximum; "+" is algebraic summation; " " is algebraic
substraction.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy relations and fuzzy
implications...
Figure 3.16
(a) Membership
functions for fuzzy
sets for the Smoker
and the Risk of
Cancer case
example.
(b) The Rc
implication relation:
"heavy smoker >
high risk of cancer"
in a matrix form.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy relations and fuzzy
implications...
Figure 3.17
Max min composition
applied over the fuzzy
set "moderate
smoker" and the Rc
implication "heavy
smoker >high risk of
cancer" from figure
3.16 for the Smoker
and the Risk of Cancer
case example.
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy relations and fuzzy
implications...
see figure 3.18 (not shown)
for examples of fuzzy compositions
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy propositions
and fuzzy logic
Fuzzy propositions are propositions which contain fuzzy
variables with their fuzzy values. The truth value of a fuzzy
proposition " X is A " is given by the membership function
mA.
Fuzzy propositions may include modifiers, or also called -
hedges. The general form of a fuzzy proposition with a
modifier is X is mA. The negation "not" can be viewed as a
modifier. Other modifiers are: very A, denoted as A2
(concentration); more or less A, denoted as A1/2 (dilation)
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy propositions and
fuzzy logic...
Figure 3.5
Five operations with two fuzzy sets A and B
approximately represented in a graphical form
© N. Kasabov Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, MIT Press, 1996
Fuzzy propositions
and fuzzy logic...
Fuzzy logic
• fuzzy propositions
• fuzzy connectives
• fuzzy rules (laws) of inference
Fuzzy laws of inference:
• generalised modus ponens
• generalised modus tolens
• de Morgan
see Figure 3.20
for an illustration of properties of fuzzy implications when used
for inference.