2008, Vol. XV, núm. 1
http://hdl.handle.net/2099/13076
Sun, 16 Jan 2022 19:28:29 GMT2022-01-16T19:28:29ZRobust rank correlation coefficients on the basis of fuzzy
http://hdl.handle.net/2099/13196
Robust rank correlation coefficients on the basis of fuzzy
Bodenhofer, U.; Klawonn, F.
The goal of this paper is to demonstrate that established rank correlation
measures are not ideally suited for measuring rank correlation for numerical
data that are perturbed by noise. We propose to use robust rank correlation
measures based on fuzzy orderings. We demonstrate that the new measures
overcome the robustness problems of existing rank correlation coe cients. As
a rst step, this is accomplished by illustrative examples. The paper closes
with an outlook on future research and applications
Mon, 15 Apr 2013 17:08:59 GMThttp://hdl.handle.net/2099/131962013-04-15T17:08:59ZBodenhofer, U.Klawonn, F.The goal of this paper is to demonstrate that established rank correlation
measures are not ideally suited for measuring rank correlation for numerical
data that are perturbed by noise. We propose to use robust rank correlation
measures based on fuzzy orderings. We demonstrate that the new measures
overcome the robustness problems of existing rank correlation coe cients. As
a rst step, this is accomplished by illustrative examples. The paper closes
with an outlook on future research and applicationsOptimization of fuzzy rule sets using a bacterial evolutionary algorithm
http://hdl.handle.net/2099/13175
Optimization of fuzzy rule sets using a bacterial evolutionary algorithm
Drobics, Mario; Botzheim, J.
In this paper we present a novel approach where we rst create a large
set of (possibly) redundant rules using inductive rule learning and where we
use a bacterial evolutionary algorithm to identify the best subset of rules in a
subsequent step. This enables us to nd an optimal rule set with respect to a
freely de nable global goal function, which gives us the possibility to integrate
interpretability related quality criteria explicitly in the goal function and to
consider the interplay of the overlapping fuzzy rules
Fri, 12 Apr 2013 19:10:10 GMThttp://hdl.handle.net/2099/131752013-04-12T19:10:10ZDrobics, MarioBotzheim, J.In this paper we present a novel approach where we rst create a large
set of (possibly) redundant rules using inductive rule learning and where we
use a bacterial evolutionary algorithm to identify the best subset of rules in a
subsequent step. This enables us to nd an optimal rule set with respect to a
freely de nable global goal function, which gives us the possibility to integrate
interpretability related quality criteria explicitly in the goal function and to
consider the interplay of the overlapping fuzzy rulesFuzzy sequential pattern mining in incomplete databases
http://hdl.handle.net/2099/13166
Fuzzy sequential pattern mining in incomplete databases
Fiot, Céline; Laurent, Anne; Teisseire, Maguelonne
Recent widening of data mining application areas have lead to new issues.
For instance, frequent sequence discovery techniques that were developed for
customer behaviour analysis are now applied to analyse industrial or biological
databases. Thus frequent sequence mining algorithm must be adapted
to handle particular characteristics of these data. Among these specificities
one should consider numerical attributes and incomplete data. In this paper,
we propose a method for discovering crisp or fuzzy sequential patterns from
an incomplete database. This approach uses partial information contained in
incomplete records, only temporary discarding the missing part of the record.
Experiments run on various synthetic datasets show the validity of our proposal
as well in terms of quality as in terms of the robustness to the rate of
missing values.
Wed, 10 Apr 2013 18:03:23 GMThttp://hdl.handle.net/2099/131662013-04-10T18:03:23ZFiot, CélineLaurent, AnneTeisseire, MaguelonneRecent widening of data mining application areas have lead to new issues.
For instance, frequent sequence discovery techniques that were developed for
customer behaviour analysis are now applied to analyse industrial or biological
databases. Thus frequent sequence mining algorithm must be adapted
to handle particular characteristics of these data. Among these specificities
one should consider numerical attributes and incomplete data. In this paper,
we propose a method for discovering crisp or fuzzy sequential patterns from
an incomplete database. This approach uses partial information contained in
incomplete records, only temporary discarding the missing part of the record.
Experiments run on various synthetic datasets show the validity of our proposal
as well in terms of quality as in terms of the robustness to the rate of
missing values.Automatic video annotation with forests of fuzzy decision trees
http://hdl.handle.net/2099/13165
Automatic video annotation with forests of fuzzy decision trees
Detyniecki, Marcin; Marsala, Christophe
Nowadays, the annotation of videos with high-level semantic concepts or
features is a great challenge. In this paper, this problem is tackled by learning,
by means of Fuzzy Decision Trees (FDT), automatic rules based on a limited
set of examples. Rules intended, in an exploitation step, to reduce the need of
human usage in the process of indexation. However, when addressing large,
unbalanced, multiclass example sets, a single classi er - such as the FDT -
is insu cient. Therefore we introduce the use of forests of fuzzy decision
trees (FFDT) and we highlight: (a) its e ectiveness on a high level feature
detection task, compared to other competitive systems and (b) the e ect on
performance from the number of classi ers point of view. Moreover, since the
resulting indexes are, by their nature, to be used in a retrieval application, we
discuss the results under the lights of a ranking (vs. a classi cation) context.
Wed, 10 Apr 2013 17:25:54 GMThttp://hdl.handle.net/2099/131652013-04-10T17:25:54ZDetyniecki, MarcinMarsala, ChristopheNowadays, the annotation of videos with high-level semantic concepts or
features is a great challenge. In this paper, this problem is tackled by learning,
by means of Fuzzy Decision Trees (FDT), automatic rules based on a limited
set of examples. Rules intended, in an exploitation step, to reduce the need of
human usage in the process of indexation. However, when addressing large,
unbalanced, multiclass example sets, a single classi er - such as the FDT -
is insu cient. Therefore we introduce the use of forests of fuzzy decision
trees (FFDT) and we highlight: (a) its e ectiveness on a high level feature
detection task, compared to other competitive systems and (b) the e ect on
performance from the number of classi ers point of view. Moreover, since the
resulting indexes are, by their nature, to be used in a retrieval application, we
discuss the results under the lights of a ranking (vs. a classi cation) context.Semigroups of semi-copulas and evolution of dependence at increase of age
http://hdl.handle.net/2099/13164
Semigroups of semi-copulas and evolution of dependence at increase of age
Foschi, R.; Spizzichino, Fabio
We consider a pair of exchangeable lifetimes
X; Y and the families of the
conditional survival functions
F t(x;y) of (X t; Y t) given (X > t; Y > t).We analyze some properties of dependence and of ageing for
F t(x;y) andsome relations among them
) of (
X
t; Y
t
) given (
X > t; Y > t
).
We analyze some properties of dependence and of ageing for
F
t
(
x; y
) and
some relations among them.
Wed, 10 Apr 2013 15:53:14 GMThttp://hdl.handle.net/2099/131642013-04-10T15:53:14ZFoschi, R.Spizzichino, FabioWe consider a pair of exchangeable lifetimes
X; Y and the families of the
conditional survival functions
F t(x;y) of (X t; Y t) given (X > t; Y > t).We analyze some properties of dependence and of ageing for
F t(x;y) andsome relations among them
) of (
X
t; Y
t
) given (
X > t; Y > t
).
We analyze some properties of dependence and of ageing for
F
t
(
x; y
) and
some relations among them.Bounds for value at risk : the approach based on copulas with homogeneous tails
http://hdl.handle.net/2099/13163
Bounds for value at risk : the approach based on copulas with homogeneous tails
Jaworski, P.
The theory of copulas provides a useful tool for modeling dependence in
risk management. In insurance and nance, as well as in other applications,
dependence of extreme events is particularly important, hence there is a need
for the detailed study of the tail behaviour of the multivariate copulas. In
this paper we investigate the class of copulas having homogeneous lower tails.
We show that having only such information on the structure of dependence
of returns from assets is enough to get estimates on Value at Risk of the
multiasset portfolio in terms of Value at Risk of one-asset portfolios
Wed, 10 Apr 2013 15:49:07 GMThttp://hdl.handle.net/2099/131632013-04-10T15:49:07ZJaworski, P.The theory of copulas provides a useful tool for modeling dependence in
risk management. In insurance and nance, as well as in other applications,
dependence of extreme events is particularly important, hence there is a need
for the detailed study of the tail behaviour of the multivariate copulas. In
this paper we investigate the class of copulas having homogeneous lower tails.
We show that having only such information on the structure of dependence
of returns from assets is enough to get estimates on Value at Risk of the
multiasset portfolio in terms of Value at Risk of one-asset portfoliosMining gradual dependencies with variation strength
http://hdl.handle.net/2099/13157
Mining gradual dependencies with variation strength
Molina, C.; Serrano, J.M.; Sánchez, D.; Vila, M.A.
In this paper we propose a definition of gradual dependence as a fuzzy association rule. Gradual dependencies represent tendencies in the variation of the degree of fulfilment of properties in a set of objects. We define the degree of variation of a certain imprecise property for a pair of objects as the difference between their membership degrees to the fuzzy set defining the property. When considering a transaction for every pair of objects and considering items representing positive and negative variations foer each property of interest, fuzzy association rules become gradual dependencies and the accuray and support of the former can be employed to assess the corresponding dependencies. We study the new semantics and properties of the resulting fuzzy gradual dependence, and we propose a way to adapt existing fuzzy association rule mining algorithms for the new task of mining such dependencies
Wed, 10 Apr 2013 14:01:51 GMThttp://hdl.handle.net/2099/131572013-04-10T14:01:51ZMolina, C.Serrano, J.M.Sánchez, D.Vila, M.A.In this paper we propose a definition of gradual dependence as a fuzzy association rule. Gradual dependencies represent tendencies in the variation of the degree of fulfilment of properties in a set of objects. We define the degree of variation of a certain imprecise property for a pair of objects as the difference between their membership degrees to the fuzzy set defining the property. When considering a transaction for every pair of objects and considering items representing positive and negative variations foer each property of interest, fuzzy association rules become gradual dependencies and the accuray and support of the former can be employed to assess the corresponding dependencies. We study the new semantics and properties of the resulting fuzzy gradual dependence, and we propose a way to adapt existing fuzzy association rule mining algorithms for the new task of mining such dependenciesLogical aggregation based on interpolative
http://hdl.handle.net/2099/13098
Logical aggregation based on interpolative
Radojevic, Dragan
Explicit inclusion of logic in the process of aggregation (information fu-
sion) is very important in real problems from many points of view such as
adequacy and transparency. In this paper aggregation is treated as a logi-
cal and/or pseudo-logical operation based on interpolative Boolean algebra
(IBA). IBA is a real-valued ([0, 1]-valued) realization of Boolean algebra.
As classical two-valued realization of Boolean algebra is the base of classical
two-valued logic, IBA is the base of real-valued logic.
Thu, 04 Apr 2013 17:19:13 GMThttp://hdl.handle.net/2099/130982013-04-04T17:19:13ZRadojevic, DraganExplicit inclusion of logic in the process of aggregation (information fu-
sion) is very important in real problems from many points of view such as
adequacy and transparency. In this paper aggregation is treated as a logi-
cal and/or pseudo-logical operation based on interpolative Boolean algebra
(IBA). IBA is a real-valued ([0, 1]-valued) realization of Boolean algebra.
As classical two-valued realization of Boolean algebra is the base of classical
two-valued logic, IBA is the base of real-valued logic.