Category: Uncategorized

Automatic localization and annotation of facial features using machine learning techniques

Abstract  Content-based image retrieval (CBIR) systems traditionally find images within a database that are similar to query image using
low level features, such as colour histograms. However, this requires a user to provide an image to the …

Abstract  

Content-based image retrieval (CBIR) systems traditionally find images within a database that are similar to query image using
low level features, such as colour histograms. However, this requires a user to provide an image to the system. It is easier
for a user to query the CBIR system using search terms which requires the image content to be described by semantic labels.
However, finding a relationship between the image features and semantic labels is a challenging problem to solve. This paper
aims to discover semantic labels for facial features for use in a face image retrieval system. Face image retrieval traditionally
uses global face-image information to determine similarity between images. However little has been done in the field of face
image retrieval to use local face-features and semantic labelling. Our work aims to develop a clustering method for the discovery
of semantic labels of face-features. We also present a machine learning based face-feature localization mechanism which we
show has promise in providing accurate localization.

  • Content Type Journal Article
  • Pages 1-15
  • DOI 10.1007/s00500-010-0586-y
  • Authors
    • Paul C. Conilione, La Trobe University Department of Computer Science and Computer Engineering Melbourne VIC 3086 Australia
    • Dianhui Wang, La Trobe University Department of Computer Science and Computer Engineering Melbourne VIC 3086 Australia

Membership evaluation and feature selection for fuzzy support vector machine based on fuzzy rough sets

Abstract  A fuzzy support vector machine (FSVM) is an improvement in SVMs for dealing with data sets with outliers. In FSVM, a key step
is to compute the membership for every training sample. Existing approaches of computing the membership o…

Abstract  

A fuzzy support vector machine (FSVM) is an improvement in SVMs for dealing with data sets with outliers. In FSVM, a key step
is to compute the membership for every training sample. Existing approaches of computing the membership of a sample are motivated
by the existence of outliers in data sets and do not take account of the inconsistency between conditional attributes and
decision classes. However, this kind of inconsistency can affect membership for every sample and has been considered in fuzzy
rough set theory. In this paper, we develop a new method to compute membership for FSVMs by using a Gaussian kernel-based
fuzzy rough set. Furthermore, we employ a technique of attribute reduction using Gaussian kernel-based fuzzy rough sets to
perform feature selection for FSVMs. Based on these discussions we combine the FSVMs and fuzzy rough sets methods together.
The experimental results show that the proposed approaches are feasible and effective.

  • Content Type Journal Article
  • Pages 1-10
  • DOI 10.1007/s00500-010-0577-z
  • Authors
    • Qiang He, Harbin Institute of Technology Department of Mathematics Harbin 150001 People’s Republic of China
    • Congxin Wu, Harbin Institute of Technology Department of Mathematics Harbin 150001 People’s Republic of China

Fuzzy decision tree based on fuzzy-rough technique

Abstract  Using an efficient criterion in selection of fuzzy conditional attributes (i.e. expanded attributes) is important for generation
of fuzzy decision trees. Given a fuzzy information system (FIS), fuzzy conditional attributes play a c…

Abstract  

Using an efficient criterion in selection of fuzzy conditional attributes (i.e. expanded attributes) is important for generation
of fuzzy decision trees. Given a fuzzy information system (FIS), fuzzy conditional attributes play a crucial role in fuzzy
decision making. Besides, different fuzzy conditional attributes have different influences on decision making, and some of
them may be more important than the others. Two well-known criteria employed to select expanded attributes are fuzzy classification
entropy and classification ambiguity, both of which essentially use the ratio of uncertainty to measure the significance of
fuzzy conditional attributes. Based on fuzzy-rough technique, this paper proposes a new criterion, in which expanded attributes
are selected by using significance of fuzzy conditional attributes with respect to fuzzy decision attributes. An illustrative
example as well as the experimental results demonstrates the effectiveness of our proposed method.

  • Content Type Journal Article
  • Pages 1-10
  • DOI 10.1007/s00500-010-0584-0
  • Authors
    • Jun-hai Zhai, Hebei University Key Laboratory of Machine Learning and Computational Intelligence, College of Mathematics and Computer Science Baoding 071002 Hebei China

DE/BBO: a hybrid differential evolution with biogeography-based optimization for global numerical optimization

Abstract  Differential evolution (DE) is a fast and robust evolutionary algorithm for global optimization. It has been widely used in
many areas. Biogeography-based optimization (BBO) is a new biogeography inspired algorithm. It mainly uses …

Abstract  

Differential evolution (DE) is a fast and robust evolutionary algorithm for global optimization. It has been widely used in
many areas. Biogeography-based optimization (BBO) is a new biogeography inspired algorithm. It mainly uses the biogeography-based
migration operator to share the information among solutions. In this paper, we propose a hybrid DE with BBO, namely DE/BBO,
for the global numerical optimization problem. DE/BBO combines the exploration of DE with the exploitation of BBO effectively,
and hence it can generate the promising candidate solutions. To verify the performance of our proposed DE/BBO, 23 benchmark
functions with a wide range of dimensions and diverse complexities are employed. Experimental results indicate that our approach
is effective and efficient. Compared with other state-of-the-art DE approaches, DE/BBO performs better, or at least comparably,
in terms of the quality of the final solutions and the convergence rate. In addition, the influence of the population size,
dimensionality, different mutation schemes, and the self-adaptive control parameters of DE are also studied.

  • Content Type Journal Article
  • Pages 1-21
  • DOI 10.1007/s00500-010-0591-1
  • Authors
    • Wenyin Gong, China University of Geosciences School of Computer Science Wuhan People’s Republic of China
    • Zhihua Cai, China University of Geosciences School of Computer Science Wuhan People’s Republic of China
    • Charles X. Ling, The University of Western Ontario Department of Computer Science London Canada

Efficient approaches for summarizing subspace clusters into k representatives

Abstract  A major challenge in subspace clustering is that subspace clustering may generate an explosive number of clusters with high
computational complexity, which severely restricts the usage of subspace clustering. The problem gets even …

Abstract  

A major challenge in subspace clustering is that subspace clustering may generate an explosive number of clusters with high
computational complexity, which severely restricts the usage of subspace clustering. The problem gets even worse with the
increase of the data’s dimensionality. In this paper, we propose to summarize the set of subspace clusters into k representative clusters to alleviate the problem. Typically, subspace clusters can be clustered further into k groups, and the set of representative clusters can be selected from each group. In such a way, only the most representative
subspace clusters will be returned to user. Unfortunately, when the size of the set of representative clusters is specified,
the problem of finding the optimal set is NP-hard. To solve this problem efficiently, we present two approximate methods:
PCoC and HCoC. The greatest advantage of our methods is that we only need a subset of subspace clusters as the input instead
of the complete set of subspace clusters. Precisely, only the clusters in low-dimensional subspaces are computed and assembled
into representative clusters in high-dimensional subspaces. The approximate results can be found in polynomial time. Our performance
study shows both the effectiveness and efficiency of these methods.

  • Content Type Journal Article
  • Pages 1-9
  • DOI 10.1007/s00500-010-0552-8
  • Authors
    • Guanhua Chen, Peking University School of Electronics Engineering and Computer Science Beijing 100871 China
    • Xiuli Ma, Peking University School of Electronics Engineering and Computer Science Beijing 100871 China
    • Dongqing Yang, Peking University School of Electronics Engineering and Computer Science Beijing 100871 China
    • Shiwei Tang, Peking University School of Electronics Engineering and Computer Science Beijing 100871 China
    • Meng Shuai, Peking University School of Electronics Engineering and Computer Science Beijing 100871 China
    • Kunqing Xie, Peking University School of Electronics Engineering and Computer Science Beijing 100871 China

The smooth switching of double modes fuzzy control

Abstract  Various multimode controls are more and more widely applied in industry to improve the performance of control systems. Double
modes fuzzy control is one of multimode controls, which has two independent and different mode controller…

Abstract  

Various multimode controls are more and more widely applied in industry to improve the performance of control systems. Double
modes fuzzy control is one of multimode controls, which has two independent and different mode controllers to satisfy different
control demands. The smooth switching of different controllers is the key technology in industrial application of multimode
modes control. Double modes fuzzy control is used to improve the dynamic and steady-state performances of control systems.
This paper focuses on the unsteady problem at switching point of controllers in double modes control system. Three structures
of double modes fuzzy control systems are proposed and discussed. The design principles of multimode control are analyzed.
Three different switching methods are analyzed and their feasibility is studied. The concept of smooth switching from one
controller to another controller is proposed. Especially the smooth switching of fuzzy/PI double modes control is analyzed,
and the corresponding fuzzy controller is designed. The simulation of smooth switching at natural switching point of fuzzy/PI
double modes control system is carried out in order to prove the superiority of smooth switching at natural switching point.
The results of this paper can offer effective reference for other multimode control design.

  • Content Type Journal Article
  • Pages 1-8
  • DOI 10.1007/s00500-010-0550-x
  • Authors
    • Fang He, Harbin Institute of Technology School of Electrical Engineering and Automation Harbin China
    • Weiming Tong, Harbin Institute of Technology School of Electrical Engineering and Automation Harbin China
    • Shouhua Zhao, University of Jinan School of Control Science and Engineering Jinan China
    • Qiang Wang, University of Jinan School of Mechanical Engineering Jinan China

-DANTE: an ant colony oriented depth search procedure

Abstract  The

e
-Depth ANT Explorer (

e
–
DANTE
) algorithm applied to a multiple objective optimization problem is presented in this paper. This method is a hybridization
of the ant colony optimization algorithm with a depth search pro…

Abstract  

The

e

-Depth ANT Explorer (

e

–
DANTE
) algorithm applied to a multiple objective optimization problem is presented in this paper. This method is a hybridization
of the ant colony optimization algorithm with a depth search procedure, putting together an oriented/limited depth search.
A particular design of the pheromone set of rules is suggested for these kinds of optimization problems, which are an adaptation
of the single objective case. Six versions with incremental features are presented as an evolutive path, beginning in a single
colony approach, where no depth search is applied, to the final

e

–
DANTE
. Versions are compared among themselves in a set of instances of the multiple objective Traveling Salesman Problem. Finally,
our best version of

e

–
DANTE
 is compared with several established heuristics in the field showing some promising results.

  • Content Type Journal Article
  • Pages 1-34
  • DOI 10.1007/s00500-010-0543-9
  • Authors
    • Pedro Cardoso, University of Algarve, ISE Campus da Penha 8005-139 Faro Portugal
    • Mário Jesus, University of Algarve, ISE Campus da Penha 8005-139 Faro Portugal
    • Alberto Márquez, University of Sevilla ESTII Avda. Reina Mercedes 41012 Sevilla Spain

Extension of the ELECTRE method based on interval-valued fuzzy sets

Abstract  Decision-making is the process of finding the best option among the feasible alternatives. In classical multiple criteria
decision-making (MCDM) methods, the ratings and the weights of the criteria are known precisely. However, if …

Abstract  

Decision-making is the process of finding the best option among the feasible alternatives. In classical multiple criteria
decision-making (MCDM) methods, the ratings and the weights of the criteria are known precisely. However, if decision makers
cannot reach an agreement on the method of defining linguistic variables based on the fuzzy sets, the interval-valued fuzzy
set theory can provide a more accurate modeling. In this paper, the interval-valued fuzzy ELECTRE method is presented aiming
at solving MCDM problems in which the weights of criteria are unequal, using interval-valued fuzzy set concepts. For the purpose
of proving the validity of the proposed model, we present a numerical example and build a practical maintenance strategy selection
problem.

  • Content Type Journal Article
  • Pages 1-11
  • DOI 10.1007/s00500-010-0563-5
  • Authors
    • Behnam Vahdani, Islamic Azad University Qazvin Branch Department of Industrial and Mechanical Engineering Qazvin Iran
    • Hasan Hadipour, Islamic Azad University Qazvin Branch Department of Industrial and Mechanical Engineering Qazvin Iran

Solving multiple instances at once: the role of search and adaptation

Abstract  Having in mind the idea that the computational effort and knowledge gained while solving a problem’s instance should be used
to solve other ones, we present a new strategy that allows to take advantage of both aspects. The strate…

Abstract  

Having in mind the idea that the computational effort and knowledge gained while solving a problem’s instance should be used
to solve other ones, we present a new strategy that allows to take advantage of both aspects. The strategy is based on a set
of operators and a basic learning process that is fed up with the information obtained while solving several instances. The
output of the learning process is an adjustment of the operators. The instances can be managed sequentially or simultaneously
by the strategy, thus varying the information available for the learning process. The method has been tested on different
SAT instance classes and the results confirm that (a) the usefulness of the learning process and (b) that embedding problem
specific algorithms into our strategy, instances can be solved faster than applying these algorithms instance by instance.

  • Content Type Journal Article
  • Pages 1-18
  • DOI 10.1007/s00500-010-0564-4
  • Authors
    • Antonio D. Masegosa, University of Granada Models of Decision and Optimization Research Group, Department of Computer Science and AI 18071 Granada Spain
    • David A. Pelta, University of Granada Models of Decision and Optimization Research Group, Department of Computer Science and AI 18071 Granada Spain
    • Juan R. González, University of Granada Models of Decision and Optimization Research Group, Department of Computer Science and AI 18071 Granada Spain

Hybrid differential evolution and Nelder–Mead algorithm with re-optimization

Abstract  Nonlinear optimization algorithms could be divided into local exploitation methods such as Nelder–Mead (NM) algorithm and
global exploration ones, such as differential evolution (DE). The former searches fast yet could be easily …

Abstract  

Nonlinear optimization algorithms could be divided into local exploitation methods such as Nelder–Mead (NM) algorithm and
global exploration ones, such as differential evolution (DE). The former searches fast yet could be easily trapped by local
optimum, whereas the latter possesses better convergence quality. This paper proposes hybrid differential evolution and NM
algorithm with re-optimization, called as DE-NMR. At first a modified NM, called NMR is presented. It re-optimizes from the
optimum point at the first time and thus being able to jump out of local optimum, exhibits better properties than NM. Then,
NMR is combined with DE. To deal with equal constraints, adaptive penalty function method is adopted in DE-NMR, which relaxes
equal constraints into unequal constrained functions with an adaptive relaxation parameter that varies with iteration. Benchmark
optimization problems as well as engineering design problems are used to experiment the performance of DE-NMR, with the number
of function evaluation times being employed as the main index of measuring convergence speed, and objective function values
as the main index of optimum’s quality. Non-parametric tests are employed in comparing results with other global optimization
algorithms. Results illustrate the fast convergence speed of DE-NMR.

  • Content Type Journal Article
  • Pages 1-14
  • DOI 10.1007/s00500-010-0566-2
  • Authors
    • Zhenxiao Gao, Tsinghua University Department of Automation Beijing China
    • Tianyuan Xiao, Tsinghua University Department of Automation Beijing China
    • Wenhui Fan, Tsinghua University Department of Automation Beijing China