Prog Artif Intell (2016) 5:1–14
DOI 10.1007/s13748-015-0069-6
REGULAR PAPER
On two construction methods of copulas from fuzzy implication
functions
Sebastia Massanet
1
· Daniel Ruiz-Aguilera
1
· Joan Torrens
1
Received: 20 November 2015 / Accepted: 23 November 2015 / Published online: 14 December 2015
© Springer-Verlag Berlin Heidelberg 2015
Abstract Copulas have been deeply investigated because
of their applications in many fields. From the theoretical point
of view, a key point in this research lies in the search of
new construction methods of parametrized families of cop-
ulas. This paper presents some construction methods based
on fuzzy implication functions by reversing the construc-
tion methods of fuzzy implication functions from copulas
presented by P. Grzegorzewski in some recent papers. Specif-
ically, the PSI and SSI-construction methods of copulas are
proposed which provide copulas from a given fuzzy implica-
tion function. In addition, the analysis of these construction
methods of copulas lead to the characterization of the inter-
section of the probabilistic S and survival S-implications with
( S, N ) and R-implications.
Keywords Fuzzy implication function · Copula ·
t-conorm · Survival copula
1 Introduction
Copulas [27] are a special kind of aggregation functions
that come from probability theory and statistics and have
important applications not only in these fields, but also in
economics mainly. The study of copulas has grown in last
B Sebastia Massanet
s.massanet@uib.es
Daniel Ruiz-Aguilera
daniel.ruiz@uib.es
Joan Torrens
jts224@uib.es
1
Department of Mathematics and Computer Science,
University of the Balearic Islands, Crta. Valldemossa km. 7.5,
Palma, Spain
decades and one of the main topics in this sense lies in
the research of different construction methods of copulas or
parametrized families of copulas. On the other hand, fuzzy
implication functions [4, 7] are a kind of logical operators that
come from fuzzy sets and fuzzy logic and they have a lot of
applications that vary from approximate reasoning and fuzzy
control [17, 22] to fuzzy subset-hood measures and image
processing [9, 10], specially in fuzzy mathematical morphol-
ogy [16].
In spite of their very different origins, both kinds of opera-
tors are highly related between them. Indeed, there are many
published works dealing with the relationship among aggre-
gation functions in general (including copulas) and fuzzy
implication functions (see [4, 8, 23, 28] and references there-
in). In particular, many authors have investigated methods
to construct fuzzy implication functions from aggregation
functions and vice versa. A well-known example of these
methods comes from the family of residual implications or
R-implications in short. Indeed, one of the earliest meth-
ods for obtaining implications was from conjunctions as
their residuals, and many different classes of conjunctions
have been used in this line. Mainly, left-continuous t-norms
[23], left-continuous conjunctive uninorms (leading to the so-
called RU -implications) [1, 12], but also semi-copulas and
copulas [14] and many other types of conjunctive aggrega-
tion functions (see [26] and references therein). In the major
part of these cases, it is also possible to construct the initial
conjunction from the corresponding residual implication.
Another construction method comes from the material
implications. In this case, given a negation N , it is possi-
ble to construct an implication function from a t-conorm
S, obtaining the well-known ( S, N )-implications, and vice
versa [4, 5]. Again, one can use a disjunctive uninorm instead
of a t-conorm (obtaining the so-called (U, N )-implications)
[6], and also many other kinds of disjunctive aggregation
123