Absolutely Monotone Real Set Functions
Biljana Mihailovi´ c
Faculty of Engineering,
University of Novi Sad
Trg Dositeja Obradovi´ ca 6,
21000 Novi Sad, Serbia
Email: lica@uns.ac.rs
Endre Pap
Department of Mathematics and Informatics,
University of Novi Sad
Trg Dositeja Obradovi´ ca 4,
21000 Novi Sad, Serbia
Email: pape@eunet.yu
Ljubo Nedovi´ c
Faculty of Engineering,
University of Novi Sad
Trg Dositeja Obradovi´ ca 6,
21000 Novi Sad, Serbia
Email: nljubo@uns.ac.rs
Abstract—We present a class of absolutely monotone and
signed stable set functions with m(∅) = 0, AMSS. The
representation of a set function from AMSS as a symmetric
maximum of two monotone set function is obtained. We present
three integrals of a real-valued measurable function based on
m ∈ AMSS.
Keywords: symmetric maximum, absolutely monotone set
function, sign stable set function
I. I NTRODUCTION
A generalization of the classical probability measure, a
fuzzy measure (capacity), together with fuzzy integrals, has
many applications in economics, pattern recognition, and
decision analysis [3], [4], [5], [20], [21]. It is proven in [1],
[16] that a real-valued set function m, m(∅), belongs to the
space of set functions with the bounded chain variation, BV ,
if it can be represented as a difference of two finite fuzzy
measures, vanishing at the empty set, i.e. m = ν
1
- ν
2
.
The most important integral defined with respect to a fuzzy
measure is introduced by G. Choquet in [2]. The Choquet
integral of a non-negative real-valued function f based on a
fuzzy measure m : A→ [0, ∞] is defined by
C
m
(f )=
∞
0
m({x|f (x) ≥ t})dt. (1)
The Choquet integral is defined with respect to non-
monotonic, real-valued set functions with the bounded chain
variation [14], and it is also known as the asymmetric Choquet
integral [3], [14], [16]. The asymmetric Choquet integral is
linear with respect to m.
An another important integral based on a fuzzy measure
m : A→ [0, 1], introduced in [18] by M. Sugeno, is the
Sugeno integral defined by
S
m
(f )= sup
t∈[0,1[
(t ∧ m({x|f (x) ≥ t})). (2)
The symmetric Sugeno integral of a real-valued function,
introduced in [5], is also defined with respect to a fuzzy
measure.
The Choquet-like integral related to some non-decreasing
function g : [0, 1] → [0, ∞], g(0) = 0, defined for a non-
negative function and a fuzzy measure m is given by
C
g
m
(f )= g
-1
(C
g◦m
(g ◦ f )) . (3)
This integral is introduced in [9] and it is also defined for a
real-valued function f , if for g is taken its odd extension on
the real line [9].
In [13] the authors introduced an absolutely monotone and
sign stabile set function m : A→ [-1, 1], m(∅)=0. It is
shown that m can be represented by a pseudo-difference of
two fuzzy measures. The class of such set functions is denoted
by AMSS.
The aim of this paper is to present the class AMSS, and
to propose different types of monotonic integrals based on a
set function from AMSS. Our previous papers [12], [13] are
devoted to non-monotonic integrals.
The paper is organized as follows. In Section 2 the pre-
liminary notions and definitions are given. In Section 3 we
consider the class of absolutely monotone and sign stabile
set functions, AMSS. In Section 4 we propose definitions of
three integrals based on m ∈ AMSS, m(X) < 1 and present
their basic properties.
II. PRELIMINARIES
The symmetric maximum is originally introduced in [5],
[6].
Definition 1: The symmetric maximum :[-a, a]
2
→
[-a, a], a ∈ R
+
is given by
xy := sign(x + y)(|x|∨|y|).
The symmetric maximum is a commutative, non-decreasing
operation with neutral element 0 and anihilator a. It is not
associative, nor continious.
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
Figure 1. Symmetric maximum
115 978-1-4244-5349-8/09/$26.00 ©2009 IEEE