Fact-checked by Grok 2 weeks ago

Fuzzy set operations

Fuzzy set operations are the mathematical procedures that extend classical to fuzzy sets, allowing elements to belong to a set to varying degrees quantified by membership functions with values in the interval [0, 1]. Formally introduced by in , these operations provide a for modeling and processing , imprecision, and in real-world scenarios, such as linguistic concepts like "tall" or "warm" where boundaries are not sharply defined. The standard fuzzy set operations, as originally proposed by Zadeh, include the , defined by the membership function \mu_{A \cup B}(x) = \max(\mu_A(x), \mu_B(x)), which captures the broadest possible membership across two sets; the intersection, given by \mu_{A \cap B}(x) = \min(\mu_A(x), \mu_B(x)), representing the overlapping membership; and the complement, specified as \mu_{\overline{A}}(x) = 1 - \mu_A(x), inverting the degree of membership. These operations inherit many algebraic properties from crisp , including commutativity (A \cup B = B \cup A), associativity ((A \cup B) \cup C = A \cup (B \cup C)), distributivity (A \cup (B \cap C) = (A \cup B) \cap (A \cup C)), and idempotence (A \cup A = A), along with De Morgan's laws (\overline{A \cup B} = \overline{A} \cap \overline{B}). However, unlike crisp sets, fuzzy operations violate the law of excluded middle (A \cup \overline{A} may not equal the universal set) and the law of contradiction (A \cap \overline{A} may not be empty), reflecting the tolerance for partial truths. Beyond these basics, fuzzy set operations encompass additional constructs like difference (\mu_{A - B}(x) = \min(\mu_A(x), 1 - \mu_B(x))) and Cartesian product for relations (\mu_{A \times B}(x, y) = \min(\mu_A(x), \mu_B(y))), as well as linguistic modifiers or hedges such as concentration (squaring membership for "very") and dilation (square root for "somewhat"). To accommodate diverse application needs, generalized operations replace min/max with triangular norms (t-norms) for intersection (e.g., algebraic product \mu_A(x) \cdot \mu_B(x)) and t-conorms for union (e.g., probabilistic sum \mu_A(x) + \mu_B(x) - \mu_A(x) \cdot \mu_B(x)), enabling behaviors ranging from conjunctive to disjunctive while preserving monotonicity and boundary conditions. Fuzzy set operations underpin broader fields like systems, approximate reasoning, and , facilitating applications in (e.g., fuzzy controllers), decision support, and where probabilistic models fall short in capturing linguistic .

Fundamentals of fuzzy sets and operations

Definition of fuzzy sets

A provides a mathematical framework for representing and by allowing elements to have degrees of membership rather than strict inclusion or exclusion. Introduced by in , extend classical to model concepts where boundaries are not sharply defined, such as linguistic terms like "approximately" or "somewhat." This approach addresses the limitations of traditional sets in handling imprecise or ambiguous information prevalent in and human reasoning. Mathematically, a A on a of X is defined as a \mu_A: X \to [0,1], where \mu_A(x) denotes the of membership of an element x \in X in A, with values ranging continuously from 0 (no membership) to 1 (full membership). This membership characterizes the by assigning grades that reflect the extent to which x belongs to A, enabling a spectrum of partial belongings. In contrast, classical crisp sets rely on a where membership is either 0 or 1, enforcing rigid dichotomies that fail to capture gradations inherent in many real-world phenomena. For instance, consider the fuzzy set of "tall people" defined over the of heights. The membership \mu_{\text{tall}}(h) might be 0 for heights below 150 cm, increase gradually to 1 for heights around 190 cm, and remain 1 thereafter, illustrating how fuzzy sets model subjective and continuous criteria. Zadeh's formulation thus laid the groundwork for operations on such sets, though the focus here remains on their definitional structure.

Motivation for fuzzy operations

Classical , grounded in two-valued , posits that an element either fully belongs to a set (membership grade 1) or does not (membership grade 0), providing a precise but rigid framework for mathematical modeling. This binary approach, however, inadequately captures the inherent vagueness and partial truths encountered in many real-world scenarios, such as linguistic descriptions in , subjective judgments in processes, and adaptive responses in systems where boundaries are imprecise or context-dependent. For instance, terms like "tall" or "hot" do not lend themselves to sharp cutoffs, leading to limitations in applying classical operations like and to phenomena involving degrees of applicability. To overcome these shortcomings, fuzzy set operations extend classical by incorporating graded memberships, as introduced by in his foundational work. Central to this extension is Zadeh's extension principle, which generalizes crisp functions to fuzzy sets by defining the membership function of the resulting fuzzy set through the supremum of input memberships mapping to the output. Specifically, for a f: X \to Y and a A on X with membership function \mu_A, the membership function of the fuzzy image f(A) is given by \mu_{f(A)}(y) = \sup_{x \in f^{-1}(y)} \mu_A(x), where f^{-1}(y) = \{x \in X \mid f(x) = y\}, and the supremum is 0 if the preimage is empty. This principle ensures that operations on fuzzy sets preserve the continuum of membership grades, allowing for a more flexible representation of uncertainty and imprecision. The development of fuzzy operations has broad implications across disciplines, serving as a cornerstone for fuzzy logic systems that emulate human reasoning, artificial intelligence applications handling ambiguous data, and pattern recognition tasks involving noisy or incomplete information. By enabling computations that reflect degrees of truth rather than absolutes, these operations facilitate robust models in environments where classical logic falters. A key benefit is the support for graded reasoning; for example, intersecting fuzzy sets representing "somewhat tall" and "somewhat heavy" can yield a fuzzy set for "somewhat overweight," capturing nuanced relationships without forcing binary classifications.

Fuzzy complements

Standard complement functions

The standard complement of a fuzzy set A, denoted A^c, is defined by the membership function \mu_{A^c}(x) = 1 - \mu_A(x) for all x in the universe of discourse. This formulation, introduced by in his seminal work on , provides the simplest algebraic representation of negation in , where the degree of membership in the complement directly inverts the original membership value on the unit interval [0,1]. Intuitively, this complement captures the notion of "not A" by assigning full membership (1) to elements completely outside A (where \mu_A(x) = 0) and zero membership to elements fully inside A (where \mu_A(x) = 1). For intermediate degrees, it linearly scales the , ensuring a straightforward probabilistic akin to classical set complements but extended to partial belongings. This makes it the most commonly used complement in basic applications, such as and control systems. Consider a fuzzy set representing "hot" temperatures, where \mu_{\text{hot}}(30^\circ \text{C}) = 0.8. The complement "not hot" then has \mu_{\text{not hot}}(30^\circ \text{C}) = 1 - 0.8 = 0.2, indicating a low but non-zero degree of membership in the cooler category. This example illustrates how the standard complement preserves the fuzzy nature while providing an intuitive duality. The standard complement exhibits key properties such as being strictly monotonic decreasing and satisfying boundary conditions \mu_{A^c}(0) = 1 and \mu_{A^c}(1) = 0, which align with the intuitive requirements for (formal axiomatic details are covered separately). It also serves as the dual operation to fuzzy intersections under in many frameworks.

Axiomatic properties of complements

Fuzzy complements in fuzzy set theory are characterized by a set of axioms that ensure their logical of classical set complements, preserving essential properties while accommodating graded membership degrees. These axioms form the foundation for defining valid complement operations and are crucial for maintaining consistency in fuzzy logical systems. The boundary conditions constitute the primary , requiring that the complement c: [0,1] \to [0,1] satisfies c(0) = 1 and c(1) = 0. This ensures that full non-membership maps to full membership in the complement and vice versa, aligning with the crisp set behavior where the complement of the is the universal set. Monotonicity requires the complement to be decreasing: if a \geq b, then c(a) \leq c(b). A stricter version, known as the strictness axiom, demands c(a) > c(b) whenever a < b, which prevents constant segments and ensures a more sensitive response to changes in membership degrees. These monotonicity properties guarantee that increasing membership in a set correspondingly decreases membership in its complement. Continuity is frequently imposed as an additional axiom, stipulating that c is a continuous function on [0,1]. This property promotes smooth transitions in fuzzy computations, which is particularly beneficial in applications requiring differentiable operations, such as optimization and control systems. The standard complement c(a) = 1 - a satisfies the boundary conditions, monotonicity (both non-strict and strict), and , serving as the baseline for fuzzy negation. A non-standard example from Yager's family is c_\omega(a) = (1 - a^\omega)^{1/\omega} for \omega > 0; this construction also adheres to the boundary conditions, strict monotonicity, and , allowing parameterization to adjust the complement's —for instance, as \omega \to \infty, it approaches 1 for a \in [0,1) and 0 for a = 1, creating a step function at the boundary a = 1.

Fuzzy intersections

Common intersection operators

Fuzzy intersection operators model the "and" relationship between fuzzy sets by combining their membership degrees in a way that generalizes classical set intersection. These operators are typically drawn from the family of triangular norms (t-norms), which provide a mathematical foundation for conjunction in fuzzy logic. Among the most widely adopted are the minimum operator, the algebraic product, and the bounded product (also known as the Łukasiewicz t-norm), each offering distinct interpretations suitable for different applications in uncertainty modeling.90241-X) The minimum operator, often referred to as the Gödel t-norm, defines the membership function of the as \mu_{A \cap B}(x) = \min(\mu_A(x), \mu_B(x)). This operator, introduced as the standard fuzzy intersection, emphasizes the limiting effect of the smaller membership degree, preserving the strongest constraint from the two sets.90241-X) It is particularly useful in scenarios where the intersection should not exceed the minimum possibility, such as in under partial . The algebraic product operator interprets intersection probabilistically, defining \mu_{A \cap B}(x) = \mu_A(x) \cdot \mu_B(x). Proposed alongside the minimum in early fuzzy set theory, this operator assumes independence between the sets and models joint occurrence akin to independent events in probability theory.90241-X) It tends to produce smaller membership values than the minimum, making it suitable for applications requiring multiplicative aggregation, like risk assessment. The bounded product, or Łukasiewicz t-norm, is given by \mu_{A \cap B}(x) = \max(0, \mu_A(x) + \mu_B(x) - 1). This operator, adapted from many-valued logic to fuzzy sets, captures a form of "cautious" conjunction where the result is zero if the sum of memberships falls below 1, reflecting strong overlap requirements. It is commonly used in fuzzy control systems and reasoning under strict conditions. To illustrate these operators, consider two fuzzy sets A and B with membership values μ_A(x) = 0.7 and μ_B(x) = 0.4 for some element x:
OperatorFormula ApplicationResult
Minimummin(0.7, 0.4)0.4
Algebraic Product0.7 × 0.40.28
Bounded Productmax(0, 0.7 + 0.4 - 1) = max(0, 0.1)0.1
These examples highlight how the minimum yields the highest intersection value, while the bounded product produces the lowest, emphasizing different degrees of conservatism.90241-X) All three operators share key algebraic properties, including commutativity—μ_{A ∩ B}(x) = μ_{B ∩ A}(x)—and associativity—(A ∩ B) ∩ C = A ∩ (B ∩ C)—ensuring consistent multi-set combinations regardless of grouping or order. These properties make them robust building blocks for complex fuzzy operations in fields like and .

Axiomatic characterization of intersections

Fuzzy intersections in fuzzy set theory are rigorously characterized through the axiomatic framework of triangular norms, or t-norms, which provide a for modeling the "and" operation between fuzzy sets on the unit interval [0,1]. This characterization ensures that the intersection operator behaves consistently with the properties expected of a in , extending classical set intersection to handle degrees of membership. The concept of t-norms originated in the work on probabilistic metric spaces and was later adapted to fuzzy sets, where it serves as the standard for defining intersections that preserve essential algebraic properties. A t-norm T: [0,1]^2 \to [0,1], denoted here as i(a,b) for fuzzy intersection, must satisfy four fundamental axioms to qualify as a valid characterization of fuzzy intersections. First, commutativity requires that i(a,b) = i(b,a) for all a,b \in [0,1], ensuring the operation is symmetric regardless of argument order. This axiom aligns with the intuitive symmetry of set intersection in classical logic and was formalized in the foundational axiomatization of associative functions. Second, monotonicity stipulates that if a \leq a' and b \leq b', then i(a,b) \leq i(a',b') for all a,a',b,b' \in [0,1], meaning the intersection is non-decreasing in both arguments. This property reflects the natural behavior where increasing membership degrees cannot decrease the intersection result, a key requirement for preserving order in fuzzy structures. The monotonicity , often stated in its one-sided form but implying the joint version, is central to the definition. Third, associativity demands that i(a, i(b,c)) = i(i(a,b), c) for all a,b,c \in [0,1], allowing the operation to be extended to multiple arguments without ambiguity, such as in i(a,b,c). This axiom enables the formation of semigroups under the operation and is crucial for applications in where chained intersections must be well-defined. Associativity traces back to early formulations in statistical triangle inequalities and is indispensable for t-norms. Finally, the boundary condition specifies that i(a,1) = a and i(a,0) = 0 for all a \in [0,1], where 1 acts as the identity element and 0 as the absorber. The identity property ensures intersection with the universal set (membership 1) yields the original set, while intersection with the empty set (membership 0) yields emptiness, mirroring classical set theory. This axiom, including the absorber derived from monotonicity, completes the t-norm structure as proposed in seminal axiomatizations. Prominent t-norms include the minimum operator i(a,b) = \min(a,b) and the product operator i(a,b) = a \cdot b, both of which satisfy the above axioms and are commonly used to verify the framework. These examples illustrate how the axiomatic properties hold for both drastic and continuous cases, with the minimum t-norm originally suggested for fuzzy intersections. In many theoretical and applied contexts, such as inference, continuity of the is additionally required to ensure measurable and differentiable operations, though not all t-norms are continuous. Dually, fuzzy unions are characterized by triangular conorms (t-conorms), which satisfy analogous axioms but with respect to the maximum and complement operations.

Fuzzy unions

Common union operators

Fuzzy union operators model the "or" relationship between fuzzy sets by generalizing the classical set union through t-conorms, which ensure boundary conditions and monotonicity. The maximum operator, also known as the Gödel t-conorm, defines the membership function of the union as \mu_{A \cup B}(x) = \max(\mu_A(x), \mu_B(x)), where \mu_A(x) and \mu_B(x) are the membership degrees of element x in fuzzy sets A and B, respectively. This operator was introduced by as the standard fuzzy union, prioritizing the higher membership degree. The probabilistic sum, or algebraic t-conorm, is given by \mu_{A \cup B}(x) = \mu_A(x) + \mu_B(x) - \mu_A(x) \cdot \mu_B(x). Proposed by Zadeh as an alternative based on probabilistic , it accounts for overlap by subtracting the product term, yielding values between the max and the simple sum. The bounded sum, known as the Łukasiewicz t-conorm, is defined as \mu_{A \cup B}(x) = \min(1, \mu_A(x) + \mu_B(x)). This , also suggested by Zadeh, caps the union at 1 to maintain membership degrees within [0,1], emphasizing additive combination until saturation. To illustrate, consider fuzzy sets with \mu_A(x) = 0.7 and \mu_B(x) = 0.4: the maximum yields 0.7, the probabilistic sum yields 0.82, and the bounded sum yields 1.0. The maximum exhibits idempotency, satisfying \max(a, a) = a for any membership degree a \in [0,1], ensuring that with itself reproduces the set.

Axiomatic characterization of unions

In fuzzy set theory, the union operation is axiomatized through the concept of triangular conorms (t-conorms), which provide a rigorous framework for generalizing the classical set union while preserving essential algebraic properties. A t-conorm S: [0,1]^2 \to [0,1] serves as the fuzzy union operator u(a,b) = S(a,b), where a and b are membership degrees in [0,1]. This characterization ensures that fuzzy unions behave intuitively as disjunctive aggregations, extending the maximum operator used in Zadeh's original formulation. The axiomatic properties defining a t-conorm are as follows:
  • Commutativity: For all a, b \in [0,1], u(a,b) = u(b,a). This symmetry reflects the indifference to order in combining fuzzy sets, mirroring the commutativity of classical s.
  • Monotonicity: For all a, a', b, b' \in [0,1] with a \leq a' and b \leq b', u(a,b) \leq u(a',b'). This non-decreasing property guarantees that increasing membership degrees does not decrease the union result, ensuring consistency with the partial order on [0,1].
  • Associativity: For all a, b, c \in [0,1], u(a, u(b,c)) = u(u(a,b), c). Associativity allows for unambiguous extension to n-ary unions, enabling the iterative combination of multiple fuzzy sets without dependency on grouping.
  • Boundary conditions: For all a \in [0,1], u(a,0) = a and u(a,1) = 1. The first condition identifies 0 as the neutral element, preserving the original membership when unioned with non-membership, while the second identifies 1 as the absorbing element, reflecting full membership dominance.
These axioms collectively characterize fuzzy unions as t-conorms, dual to the t-norms used for intersections in fuzzy set theory. Specifically, given a fuzzy complement c: [0,1] \to [0,1] (typically the standard negation c(x) = 1 - x) and a t-norm i, the dual t-conorm is defined by u(a,b) = c \bigl( i \bigl( c(a), c(b) \bigr) \bigr). This duality ensures that fuzzy unions and intersections are complementary operations, satisfying De Morgan laws under appropriate negations. A canonical example is the maximum operator, u(a,b) = \max(a,b), which arises as the dual of the minimum t-norm under the standard complement: \max(a,b) = 1 - \min(1-a, 1-b). This illustrates how the axiomatic framework encompasses Zadeh's original union while allowing for broader generalizations.

General aggregation functions

In fuzzy set theory, general aggregation functions extend the concepts of binary intersections and unions to combine multiple membership degrees into a single value, enabling the handling of multi-input scenarios. Formally, an aggregation function f: [0,1]^n \to [0,1] maps a vector of n values from the unit interval to a single output in the same interval, generalizing operations like the minimum (for conjunction-like behavior) and maximum (for disjunction-like behavior). T-norms and t-conorms represent special cases of such functions when n=2. Averaging operators form a key subclass of aggregation functions, providing compensatory mechanisms that balance inputs without extreme dominance by any single value, unlike the non-compensatory min or max. The arithmetic mean, defined as \mu = \frac{1}{n} \sum_{i=1}^n \mu_i, exemplifies this by yielding the central tendency of the inputs. Other examples include the harmonic mean, \mu = \frac{n}{\sum_{i=1}^n \frac{1}{\mu_i}} (for \mu_i > 0), which emphasizes smaller values, and the geometric mean, suitable for multiplicative aggregation. For instance, aggregating three membership degrees 0.2, 0.5, and 0.8 via the arithmetic mean results in 0.5, reflecting balanced compensation. Advanced averaging operators, such as the ordered weighted averaging (OWA) operator, introduce flexibility by assigning weights based on the ordered ranks of inputs rather than their positions. Introduced by Yager, the OWA operator is given by \text{OWA}(\mu_1, \dots, \mu_n) = \sum_{i=1}^n w_i \mu_{(i)}, where \mu_{(i)} are the sorted inputs in non-increasing order and w_i are weights summing to 1, allowing parameterization from min-like to max-like behavior. For the example values 0.2, 0.5, 0.8 with equal weights w_i = 1/3, OWA yields the of 0.5; unequal weights, such as emphasizing the largest, could produce values closer to 0.7. Many aggregation functions, including averaging operators, satisfy desirable properties such as non-decreasing monotonicity—ensuring that increasing any input does not decrease the output—and idempotency, where f(x, \dots, x) = x for all x \in [0,1], preserving consistent inputs without alteration. These properties ensure reliable behavior in applications. In multi-criteria , aggregation functions integrate diverse fuzzy evaluations from multiple criteria or experts, facilitating overall assessments in uncertain environments.

Implications and negations in fuzzy logic

In fuzzy logic, implication operators generalize the classical "if-then" relation to accommodate degrees of membership in [0,1], enabling approximate reasoning under uncertainty. These operators, denoted as I(a, b) where a and b represent truth values of antecedent and consequent respectively, must satisfy properties such as I(0, b) = 1 for all b \in [0,1] (true when antecedent is false) and I(1, b) = b (preserving the consequent when antecedent is true). Seminal classifications distinguish S-implicators, derived from t-conorms and negations, and residual implicators from t-norms, each serving distinct roles in logical inference. S-implicators, also known as (S,N)-implicators, are constructed as I(a, b) = S(N(a), b), where N is a fuzzy (typically the standard N(a) = 1 - a) and S is a t-conorm (e.g., the maximum S(a, b) = \max(a, b)). This yields the Kleene-Dienes implicator I(a, b) = \max(1 - a, b), derived from the maximum t-conorm, which satisfies the exchange principle I(a, I(a, b)) = I(a, b) and is widely used for its simplicity in modeling material implications. These operators ensure monotonicity and boundary conditions, making them suitable for probabilistic interpretations in fuzzy systems. Residual implicators, or R-implicators, arise from s T via the residuation I(a, b) = \sup\{z \in [0,1] \mid T(a, z) \leq b\}, introduced by Trillas and Valverde to model fuzzy . For the Łukasiewicz T(a, b) = \max(0, a + b - 1), this gives I(a, b) = \min(1, 1 - a + b), which is and supports the law of explosion in fuzzy settings. s from continuous t-norms are left-continuous and satisfy the adjointness property T(a, I(a, b)) \leq b \leq I(a, T(a, b)), ensuring logical in . Unlike S-implicators, R-implicators prioritize algebraic duality with conjunctions, facilitating applications in and multi-valued logics. Beyond standard complements N(a) = 1 - a, fuzzy negations include strong negations that are strict (continuous, strictly decreasing) and involutive, satisfying N(N(a)) = a for all a \in [0,1]. Such negations, parameterized by strictly increasing generators g: [0,1] \to [0,\infty) with g(0)=0, take the form N(a) = g^{-1}(1 - g(a)) and preserve double negation elimination, unlike weaker negations. The standard negation is a prototypical strong example, but others like Yager's N_\omega(a) = (1 - a^{1/\omega})^{1/\omega} for \omega > 0 allow tunable strictness, enhancing flexibility in handling linguistic hedges. These implications and negations underpin fuzzy rule-based systems, where rules of the form "if A then B" are evaluated using generalized : given input A', the output B' satisfies T(\mu_{A'}(x), I(\mu_A(x), \mu_B(y))) \leq \mu_{B'}(y). implicators excel in precise for control systems, while S-implicators support robust approximation in , as seen in Mamdani's linguistic for dynamic plants. Strong negations enable contrapositive symmetrization, I(a, b) = I(N(b), N(a)), improving completeness in systems.

References

  1. [1]
  2. [2]
    [PDF] Fuzzy Sets and Fuzzy Logic: Theory and Applications
    Klir, George J., 1932-. Fuzzy sets and fuzzy logic: theory and applications / George J. Klir, Bo Yuan. p. cm. Includes bibliographical references and index ...
  3. [3]
    Fuzzy sets - ScienceDirect.com
    June 1965, Pages 338-353 ... In particular, a separation theorem for convex fuzzy sets is proved without requiring that the fuzzy sets be disjoint.
  4. [4]
    [PDF] Fuzzy Sets* - - Delft Center for Systems and Control
    A fuzzy set is a class of objects with a continuum of grades of membership. Such a set is characterized by a membership (charac-.
  5. [5]
    Triangular norms and conorms - Scholarpedia
    Mar 15, 2007 · Triangular norms and conorms are operations which generalize the logical conjunction and logical disjunction to fuzzy logic.Missing: original paper
  6. [6]
    Triangular norms. Position paper I: basic analytical and algebraic ...
    We present the basic analytical and algebraic properties of triangular norms. We discuss continuity as well as the important classes of Archimedean, strict and ...
  7. [7]
    Triangular Norms - SpringerLink
    Free delivery 14-day returnsBerthold Schweizer and Abe Sklar in [Schweizer & Sklar 1958, 1960, 1961] provided the axioms oft-norms, as they are used today, and a redefinition of ...
  8. [8]
    [PDF] Triangular norms. Position paper I: Basic analytical and algebraic ...
    Finally we give a short historical overview on the development of t-norms and their way into fuzzy sets and fuzzy logics. To keep the paper readable, we have ...
  9. [9]
    Page not found
    Insufficient relevant content. The provided URL (https://www.cambridge.org/core/books/aggregation-functions/9780521519140) returns a "Page not found" error, and no substantive text is available for extraction or summarization.
  10. [10]
  11. [11]
  12. [12]
    Fuzzy Implications | SpringerLink
    In stock Free deliveryFuzzy Implications from Fuzzy Logic Operations. Michał Baczyński, Balasubramaniam Jayaram. Pages 39-107. Fuzzy Implications from Generator Functions. Michał ...
  13. [13]
    (PDF) Implication operators in fuzzy logic - ResearchGate
    Aug 5, 2025 · Abstract. The choice of fuzzy implication as well as other connectives is an important problem in the theoretical development of fuzzy logic ...Missing: Kleene- Dienes
  14. [14]
    [PDF] On the continuity of residuals of triangular norms
    Trillas, L. Valverde, On implication and indistinguishability in the setting of fuzzy logic, in: J. Kacprzyk, R.R. Yager (Eds.), Management Decision. Support ...
  15. [15]
    [PDF] YAGER'S CLASSES OF FUZZY IMPLICATIONS - Kybernetika
    If N is a strong negation, then I is called a strong implication (or S-implication). The family of all (S, N)-implications will be denoted by IS,N . The ...