Summary of "session 2 : classical set vs fuzzy set (fuzzylogic arabic )"

Session 2 — Classical (crisp) sets vs. Fuzzy sets (fuzzy logic) — Summary

Purpose

Review classical (crisp) set concepts and then explain how fuzzy (membership‑degree) sets differ in representation, operations, and properties. Prepare to solve exercises converting/relating T‑norms and S‑norms.

Classical (crisp) sets — brief reminder

Two standard representations:
- Extensional (list of elements): e.g., A = {Cairo, Alexandria, Tanta}
- Intensional/definitional: A = {x ∈ U | property(x)} (e.g., population > threshold)
Standard set operations: intersection, union, complement, difference.
Classical laws (examples):
- A ∩ A^c = ∅
- A ∪ A^c = U

Fuzzy (membership‑degree) sets — key differences

Elements have degrees of membership µ_A(x) ∈ [0,1] instead of binary membership {0,1}.
Two ways to present a fuzzy set:
- Extension‑style (discrete listing with membership degrees): A = x1/µ(x1) + x2/µ(x2) + … (The “+” separates element/degree pairs, not arithmetic addition.)
- Intensional‑style (membership function): specify µ_A(x) as a real‑valued function over the universe; for continuous universes use formulas, integrals or graphs.
- Common membership‑function shapes: triangular, trapezoidal, Gaussian (choose by modeling needs).

Fuzzy set operations (generalized via T‑norms and S‑norms)

Intersection: µ_{A∩B}(x) = T(µ_A(x), µ_B(x)) where T is a chosen T‑norm.
Union: µ_{A∪B}(x) = S(µ_A(x), µ_B(x)) where S is a chosen S‑norm (t‑conorm).
Complement: µ_{A^c}(x) = 1 − µ_A(x).
Common (Zadeh) choices:
- T‑norm (intersection): T_min(a,b) = min(a,b)
- S‑norm (union): S_max(a,b) = max(a,b)
Other norms exist (algebraic product, probabilistic sum, Łukasiewicz, etc.). Choice depends on the application and must satisfy certain axioms.

Axioms / requirements for T‑norms and S‑norms

To preserve sensible behavior, T‑norms (and similarly S‑norms) should satisfy:

Commutativity: T(a,b) = T(b,a)
Associativity: T(a,T(b,c)) = T(T(a,b),c)
Monotonicity: if a ≤ a' and b ≤ b' then T(a,b) ≤ T(a',b')
Boundary / identity element: T(a,1) = a (1 is identity for T) For S‑norms the identity is swapped: S(a,0) = a (0 is identity for S)

Duality (De Morgan relation) between T and S

S(a,b) = 1 − T(1 − a, 1 − b)
Conversely: T(a,b) = 1 − S(1 − a, 1 − b) Use these relations to derive an appropriate S‑norm from a given T‑norm and vice versa.

Practical computation / methodology (step‑by‑step)

Represent the fuzzy sets:
- Discrete case: list element/degree pairs, e.g., A = x1/µ_A(x1) + x2/µ_A(x2) + …
- Continuous case: specify µ_A(x) (triangular, trapezoidal, Gaussian, etc.)
To compute intersection A ∩ B:
- Choose a T‑norm (e.g., min, product, or another valid T‑norm).
- For each x compute µ_{A∩B}(x) = T(µ_A(x), µ_B(x)).
To compute union A ∪ B:
- Choose an S‑norm (e.g., max, algebraic sum) or derive it from a T‑norm via De Morgan duality.
- For each x compute µ_{A∪B}(x) = S(µ_A(x), µ_B(x)).
To compute complement A^c:
- For each x compute µ_{A^c}(x) = 1 − µ_A(x).
To convert between T‑norm and S‑norm:
- Use S(a,b) = 1 − T(1 − a, 1 − b) (and the inverse).
Implementation notes:
- Continuous representations: use integral/graphical methods.
- Discrete representations: apply operations elementwise.

Differences in logical/classical properties

In crisp sets certain identities always hold (e.g., A ∩ A^c = ∅, A ∪ A^c = U).
In fuzzy sets these equalities generally do not hold: an element may partially belong to both A and A^c; results depend on the chosen norms.

What comes next (lecturer’s plan)

Next session: solve exercises applying chosen T‑norms/S‑norms, compute intersections/unions/complements, and derive equivalent norms (T ↔ S) for given norms.

Important formulas (compact)

µ_{A∩B}(x) = T(µ_A(x), µ_B(x))
µ_{A∪B}(x) = S(µ_A(x), µ_B(x))
µ_{A^c}(x) = 1 − µ_A(x)
Duality: S(a,b) = 1 − T(1 − a, 1 − b)

Speakers / sources referenced

Lecture speaker: the instructor (unnamed)
Referenced theorists and ideas:
- Lotfi A. Zadeh — originator of fuzzy sets and the min/max (Zadeh) operators
- Aristotle — referenced regarding classical logic
- De Morgan — De Morgan’s laws / duality