Soft Computing Research

Soft Computing Developments of the Applications of Fuzzy Logic and
Evolutionary Algorithms Research Unit
at the European Centre for Soft Computing

Oscar Cordón, Sergio Damas, Raul Del Coso, Oscar Ibáñez and Carmen Peña
European Centre for Soft Computing
Edificio Científico-Tecnológico. C. Gonzalo Gutiérrez Quirós, s.n.
33600 Mieres (Asturias). Spain


This contribution is devoted to review the outcomes of some of the research lines developed at the European Centre for Soft Computing since its creation by the beginning of 2006. In particular, the activities performed by the Applications of Fuzzy Logic and Evolutionary Algorithms research unit will be described. We will specially focus on two challenging research projects, at different stages of development, which actually show the potentials of soft computing in two different medical scenarios: forensic identification and medical imaging.

1. Introduction

Soft Computing (SC) [Bonissone97] is an area of artificial intelligence research focused on the design of intelligent systems to process uncertain, imprecise and incomplete information. SC methods applied to real-world problems offer more robust, tractable and less costly solutions than those obtained by more conventional mathematical techniques [Karray04].

The main constituents of SC are fuzzy logic [Klir96], neural networks [Rumelhart86], evolutionary computing [Back97] and probabilistic reasoning [Pearl88]. Since Lotfi A. Zadeh coined the term in 1991 [Zadeh94], this technological area has developed rapidly both in its theoretical aspects and in its business applications. SC techniques address different types of problems both in typology (modeling, optimization, planning, control, forecasting, data mining, etc.) and in the areas of application (industrial production, logistics, energy, banking, food industry, etc).

One of the consequences of the significant development experienced by the SC area has been the creation of the European Centre for Soft Computing (ECSC), an international research centre specifically devoted to the topic, in Mieres (Asturias), Spain by 2006. The aim of the current contribution is to briefly describe the research lines developed and under development in the Applications of Fuzzy Logic and Evolutionary Algorithms (AFE) research unit at the ECSC. In particular, a strong focus will be put on the description of two representative research projects.

On the one hand, we will present the successful results obtained in a three-year multidisciplinary project entitled “Soft Computing and Computer Vision in Forensic Identification”, which was granted by the Spanish Ministry of Education and Research in 2006 and will be finished by the end of 2009. It was aimed to develop an intelligent system based on SC techniques to assist the forensic anthropologist in the identification of a missing person. The underlying forensic identification technique is called craniofacial superimposition and it is based on overlaying a scanned 3D model of the skull found against a subject’s face photo trying to establish whether this is the same person through the partial matching of two sets of radiometric points. We will show how evolutionary algorithms and fuzzy logic can become powerful supporting means for the forensic expert in this identification procedure by reporting its superb performance when solving some real-world identification cases from the Physical Anthropology lab at the University of Granada, Spain.

On the other hand, we will also introduce a recently launched research project entitled “Medical Imaging Using Bio-inspired and Soft Computing”, coordinated by the AFE research unit head, Dr. Cordón, and funded by the European Commission under the Marie Curie International Training Network action within the Seventh Framework Program (FP7-PEOPLE-ITN-2008). The main goal of this second project is to compose a high quality interdisciplinary training program for 16 young researchers in order they can develop their doctoral dissertations in the application of intelligent systems constituted by SC and bioinspired computing techniques to real-world medical imaging applications. To do so, a multidisciplinary partnership composed of 12 institutions (research centers, universities, companies, and hospitals) has been established including prestigious researchers in the SC-bioinspired computing (e.g., Prof. Kerre from Gent University in Belgium, Prof. Herrera from University of Granada in Spain, and Dr. Dorigo from Université Libre de Bruxelles in Belgium) and medical imaging fields (e.g., Prof. Henning from Universitätsklinikum Freiburg in Germany and Prof. Li from the University of Nottingham in the UK).

The structure of the paper is as follows. The next section will briefly introduce the ECSC by reviewing its current structure, vision, goals, scientific committee, and main research and training activities. Section 3 will focus on the AFE unit by describing its current main research lines and recent developments. Section 4 will be devoted to describe the two said research projects. Finally, some concluding remarks will be pointed out in Section 5.

2. The European Centre for Soft Computing

The ECSC is a young international research and development centre located in Mieres (Spain) with the purpose of serving as a world-class institution focused on basic and applied research in the area of SC. It was launched by the beginning of 2006, supported by a private non-profit foundation (Foundation for the Advancement of Soft Computing).

The main ECSC goal is the basic and applied research in the SC area as well as the technology transfer in industrial applications of intelligent systems design for the resolution of real-world problems. Besides, the Centre wants to be a meeting point for worldwide experts and also a place where PhD students and young researchers can develop advanced research. The official language of the ECSC is English. Since its creation, non-Spanish researchers have worked at the ECSC. Currently, approximately 40% of the researchers of the Centre are from abroad (India, Germany, Poland, Netherlands, etc.).

The motivation for creating the ECSC is based on the vision of SC research as a high potential tool for innovation and economic development. This vision is developed in a mission that involves three basic elements:

Furthermore, the mission can be summarized into four strategic lines that guide the activity of the ECSC:

In the context of these strategic lines, the main objectives of the ECSC are focused on the generation of new scientific knowledge and on the application of information technologies. These objectives include both theoretical approaches and the application of SC to solve real-world problems of industry, economy, and society. The Centre interacts with its stakeholders by cooperating with universities, research organizations and companies in R&D activities, providing specialized training and the social dissemination of research results.

In order to accomplish the previous objectives, the ECSC is comprised by five research units:

The ECSC is supported by a Scientific Committee composed of ten renowned international researchers. Its functions include the definition of the main lines of research, advice in the recruitment of top researchers, and the periodical assessment of scientific and technical performance of the Centre. In its original composition, the Scientific Committee was chaired by Lotfi A. Zadeh (Univ. Berkeley, EE.UU.) and vice-chaired by Enric Trillas (ECSC), with the support of María ángeles Gil (Univ. Oviedo, Spain) in the secretary role. It was also comprised by the following seven members: Senén Barro (Univ. Santiago de Compostela, Spain), Christer Carlsson (Univ. Åbo Akademi, Finland), Janusz Kacprzyk (Intelligent Systems Laboratory, Poland), Rudolf Kruse (Univ. Otto-von-Guericke Magdeburg, Germany), Ebrahim Mamdani (Imperial College London, UK), Henri Prade (Centre National de la Recherche Scientifique, France), and Gianguido Rizzotto (SST Group, Italy). Presently, those members are being replaced. In particular, Dr. Barro and Dr. Rizzotto have been substituted by Dr. Javier Montero (University Complutense of Madrid, Spain) and by Dr. Piero P. Bonissone (General Electrics Research, USA).

The ECSC has established numerous collaborations with the industry to apply SC techniques to improve business productivity and create new products and services. In this context, the Centre has developed and is developing around 14 R&D projects with companies including a small company in the agricultural sector, technological high-growth small and medium enterprises (SMEs), 6 large national public-private consortia, and 3 projects with large multinationals such as EDP (an energy company from Portugal), Tenneco (an American automobile components company), and the PMG Group.

Currently, the ECSC participates in 7 basic and applied research projects funded by the Government of Spain and the regional government of Asturias. In terms of EU funding, the ECSC is coordinating the Seventh Framework Program (FP7) Marie Curie Initial Training Network “Medical Imaging Using Bio-inspired and Soft Computing” (see Section 4.2) and the COST Action “Combining Soft Computing Techniques and Statistical Methods to Improve Data Analysis Solutions”. In addition, the ECSC participates in the FP7 ICT FET project “Bisociation Networks for Creative Information Discovery” and has recently received a Marie Curie International Incoming Fellowship on “Soft Collaborative Intelligent Systems”.

Furthermore, the Centre shows an active participation in SC training activities [Magdalena09]. It organizes a yearly summer course in SC that is taught by international renowned researchers. Besides, the ECSC, in collaboration with the University of Oviedo, coordinates an official Master (new European denomination for PhD program) in “Soft Computing and Intelligent Data Analysis” taught in English and adapted to the European Higher Education Area.

3. The Applications of Fuzzy Logic and Evolutionary Algorithms Research Unit

The main aim of this ECSC research unit is to propose new methodologies to tackle complex real-world problems by means of evolutionary algorithms (EAs), fuzzy logic (FL) and fuzzy systems (FSs), either in isolation or by their hybridization. Among these problems, we find those in the optimization, system identification (modeling, classification, and forecasting), data mining, and intelligent data analysis domains.

FL extends classical logic to provide a conceptual framework for knowledge representation under imprecision and the consequent uncertainty, while a FS is any kind of FL-based system using FL for knowledge representation and approximate reasoning. On the other hand, EAs are a kind of learning and optimization algorithms based on computational models of evolutionary processes.

To our mind, the combination of FL and FSs ability to model real-world phenomena presenting uncertainty and vagueness, and the search and knowledge discovery capability of EAs can be of help to solve some problems where either classical techniques can not been applied or they can be outperformed by intelligent techniques of the latter kind.

This research unit is active at the ECSC since April, 2006. It is currently composed of six researchers: Dr. Oscar Cordón (Principal researcher), Dr. Sergio Damas (Associate researcher), Dr. Arnaud Quirin (Postdoctoral researcher), Dr. Prakash Shelokar (Postdoctoral researcher), Mr. Oscar Ibáñez (Research assistant) and Mr. Krzysztof Trawinski (Research assistant). It also includes Dr, José Santamaría from the University of Jaén (Spain) and Mr. Manuel Chica from the Inspiralia Tecnologías Avanzadas company in Madrid (Spain) as external affiliated researchers.

The following four subsections are devoted to briefly describe the research lines being developed by the AFE unit, enumerating the collaborations established with other research groups. Besides, the fifth subsection will provide a list of the currently active research project and contracts.

3.1. Multi-objective Graph-based Data Mining. Design and Mining of Visual Science Maps

In spite of the fast and huge development experienced by the data mining and knowledge discovery field in the last few years, current tools and techniques to examine the content of large databases are still hampered by their inability to support searches based on criteria that are meaningful to users of those repositories. These shortcomings are particularly evident in data banks storing representations of structural objects, such as biological networks, scientific data, satellite maps, organizations’ technical documentation, semantic web, CAD circuits, and many others.

Recently, conceptual clustering techniques have been successfully applied to structural data to uncover objects or concepts that relates objects (i.e., subgraphs that represent associations of features), thus creating the area of graph-based data mining and knowledge discovery [Holder05, Washio03]. However, they still carry the pitfall that the formulation of the search problem in a graph-based structure would result in the generation of many substructures with small extent as it is easier to explain or model much smaller data subsets than those that constitute a significant portion of the dataset. For this reason, any successful methodology should also consider additional, conflicting criteria to extract better defined concepts based on the size of the substructure being explained, the number of retrieved substructures, and their diversity.

To deal with this problem, we have proposed a methodology called Evolutionary Multi-Objective Conceptual Clustering (EMO-CC) [Romero08] that uses multi-objective and multimodal EAs (specifically, genetic programming [Koza92]) to evaluate concepts or substructures based on the former conflicting criteria, and thus, to retrieve a Pareto set of non-dominated, meaningful substructures from structural databases in a single run. This approach allows us to uncover cohesive clusters comprising even an small number of observations (and not only the most frequent ones, as usual) that describe the underlying phenomena from different angles, revealing novel information that otherwise would be concealed by uninformative frequent descriptions.

EMO-CC was proposed for Bioinformatics problems but it is easily applicable to different domains by customizing the optimization objectives used to evaluate interesting substructures. Our current main application field is the mining of visual science maps (scientograms), a very novel, useful tool for the analysis of scientific information, which is built from co-citation information using classical methods from the field of bibliometrics such as citation analysis, and social networks analysis and information visualization techniques [Vargas07]. We are currently performing scientogram mining in order to analyze and compare the structure of scientific fields and research fronts in maps of the same (taken at different periods of time) or different domains (looking for similarities between different countries scientific productions) [Quirin09b] using EMO-CC and other graph-based data mining techniques such as Subdue [Cook00].

This work is done in cooperation with the Scimago research group (http://www.scimago.es/) headed by Prof. Felix de Moya at the University of Granada (Spain), which has developed two very ambitious projects called The Atlas of Science (http://www.atlasofscience.net/), to create a web-based information system achieving a graphic representation of all the IberoAmerican Science Research; and the SCImago Journal & Country Rank (http://www.scimagojr.com/), which provides new scientific journal quality indicators to assess and analyze scientific domains. Apart from the scientogram mining approach, we have developed several methods for scientogram design in the latter two web information systems, such as new network pruning algorithms significantly reducing the Pathfinder [Schvaneveldt90] run time in order to allow us to generate scientograms of very large scientific domains (even of the whole World production) in an on-line fashion [Quirin08a,08b], and new network visualization approaches achieving closer representations to human beings’ understanding [Quirin09a].

Besides, we have applied the latter methods to other domains and problems such as multi-agent systems debugging, in collaboration with Dr. Juan Botia’s research team at the University of Murcia (Spain) [Serrano10]. Finally, we are proposing new graph-based data mining methods considering multi-objective ant colony optimization algorithms [García07], well suited to explore graph structures.

3.2 Fuzzy Classifier Derivation for High Dimensional Problems with Good Interpretability-Accuracy Trade-off

System identification involves the use of mathematical tools and algorithms to build dynamical models describing the behavior of real-world systems from measured data. There are always two conflicting requirements in the modeling process: the model capability to express the behavior of the real system in an understandable way (interpretability) and its capability to faithfully represent the real system (accuracy) [Casillas03a,b]. Obtaining high degrees of interpretability and accuracy is a contradictory purpose and, in practice, one of the two properties prevails over the other.

FSs have demonstrated their outstanding capability as system identification and control tools. FL has proven its ability to generate different kinds of fuzzy models/classifiers/controllers, with a different accuracy-comprehensibility trade-off, and to permit the incorporation of human expert knowledge; as well as to integrate numerical and symbolic processing into a common scheme.

We are world-wide recognized experts on the design of FSs by means of EAs, the so called genetic fuzzy systems [Cordón01,04]. Among other real-world applications, we have used them to build fuzzy models for the estimation of maintenance costs of electricity distribution networks in Asturias [Cordón99] (outperforming other approaches such as neural networks and classical and symbolic regression), in collaboration with the head of the “Metrología y Modelos” research group at the University of Oviedo (Spain), Dr. Luciano Sánchez. Moreover, they were also applied to derive fuzzy controllers for HVAC systems for large buildings, simultaneously optimizing several design criteria [Alcalá03].

At this moment, we are developing a new approach to design ensembles of fuzzy classifiers able to deal with high dimensional problems, by considering data re-sampling and feature selection techniques, and multi-criteria EAs for the individual classifier selection to get an appropriate accuracy-interpretability trade-off [Cordón09,10, Trawinski09].

3.3. Soft Computing for Medical Image Processing

In the last few years there is an increasing interest on using SC techniques to solve real-world image processing problems covering a wide range of domains. In particular, one of the application fields that has suffered a large development is that of image registration (IR) [Goshtasby05]. IR is a fundamental task in computer vision used to achieving the best fitting/overlaying between two (or more) different images taken under different conditions (at different times, using different sensors, from different viewpoints, or a combination of them). Over the years, it has been applied to a broad range of situations from remote sensing to medical images or artificial vision and CAD systems, and different techniques have been independently studied resulting in a large body of research.

In this way, evolutionary IR is a very promising application area nowadays. Thanks to their global optimization techniques nature, EAs aim at solving the drawbacks presented by traditional IR methods, which usually get stuck in local optima when dealing with large misalignments between the images to be registered [Besl92]. Our team has developed a large number of robust evolutionary IR approaches able to overcome the latter problems based on the use of advanced EAs including domain knowledge, such as [Cordón06a,b] among others. They have achieved a successful performance on both medical IR (human MR and CT images) and on 3D model reconstruction (range images).

Specifically, we are currently extending the latter methods, hybridized with FL, to deal with a challenging real-world problem from the field of forensic medicine, in cooperation with the Physical Anthropology Lab of the University of Granada headed by the prestigious forensic anthropologist Dr. Miguel Botella. This project will be described in detail in Section 4.1. On the other hand, we also coordinate a Marie Curie International Training Network entitled “Medical Imaging Using Bio-inspired and Soft Computing” which has been recently granted by the European Commission within the Seventh Framework Program (FP7-PEOPLE-ITN-2008). A description of this project will also be provided in Section 4.2.

3.4. Real-World Applications of Single and Multi-objective Metaheuristics

Many complex combinatorial and numerical optimization problems arise in human activities, such as Economics (e.g., portfolio selection), Industry (e.g., scheduling or logistics), or Engineering (e.g., routing). The impracticability to get optimal solutions for these kinds of problems in reasonable time using classical algorithmic techniques has caused the successful development of different approximate algorithm methodologies called metaheuristics [Glover03] in the last two decades, able to quickly provide high quality solutions to them. Their success when solving a large number of real-world optimization problems is due both to the powerful heuristic search they apply in complex, ill-defined solution spaces of huge dimension, and to their flexibility, which allows them to handle problem restrictions in an easier way or to be able to simultaneously optimize multiple, conflicting objectives, which are usually present in these problems.

Metaheuristics constitute a very diverse family of optimization algorithms. Our staff owns a large expertise on the single- and multi-objective variants of many of them, mainly on EAs, ant colony optimization (ACO) [Dorigo04], scatter search, simulated annealing, tabu search, GRASP, and iterated local search. We have both used them in different applications such as medical IR, bioinformatics (genetic regulatory networks knowledge discovery), or information retrieval, as well as we designed new hybrid designs in the field of ACO aiming to obtain better performing algorithms.

Currently, we are applying multi-objective ACO algorithms [García07] to solve a challenging real-world problem, the time and space assembly line problem (TSALBP) [Bautista07], which involves to achieve optimal assignments of a subset of tasks to each station of the assembly line of a plant with respect to two or three conflicting objectives to be minimized: its cycle time, its number of stations, and their area [Chica09a,b,c]. This framework emerged thanks to the observation of a real automotive industry plant belonging to Nissan and located in Barcelona (Spain), as this research is being performed in collaboration with the Nissan Endowed Chair of the Technical University of Catalonia (http://www.nissanchair.com/), headed by Prof. Joaquín Bautista.

In the short future, we aim to use other metaheuristics to solve sales force deployment problems for an Asturian company, MOSA (http://www.mosagrupo.com/), a perfumery and food wholesaler.

3.5. Active Research Projects and Contracts

A detailed list of the currently active research projects and contracts at the AFE unit is provided as follows:

4. Two challenging projects at the AFE research unit

A brief description of two of the projects tackled by the members of the AFE research unit follows. Both projects have been selected as good representatives of the goals of this research unit. In particular, they demonstrate the possibility to apply SC to challenging problems for the society. Moreover, they are an opportunity to join the forces of world-wide recognized researchers as a step forward in the development, application and knowledge transfer of SC concepts.

4.1. SOCOVIFI: Soft Computing and Computer Vision in Forensic Identification

Our main objective is to develop an intelligent system to assist the forensic anthropologist in the identification of a missing person by a usual forensic identification technique called craniofacial superimposition [Krogman86, Iscan93]. This technique is based on overlaying a scanned 3D model of the skull found against a face photo to try to establish whether this is the same person through the partial matching of two sets of radiometric points. In order to do so, we cooperate with the Physical Anthropology Lab of the University of Granada (Spain), headed by the prestigious forensic anthropologist Dr. Miguel Botella. His team is internationally recognized by its participation in forensic anthropology activities such as the identification of Christopher Columbus’ skeletal remains, the identification of the women killed at Ciudad Juárez (Mexico), or the identification of victims of the dictatorial repression at Chile, among many others.

Our research will be based on the use of three main techniques: EAs, FL, and image processing (especially, IR). While EAs and image processing techniques have been successfully used to automatically build the skull 3D model and perform the skull-face superimposition, FL will be considered for the landmark matching and to suggest the final identification decision to the forensic expert. Specifically, we extended our previous evolutionary IR methods (see Section 3.3) for classical medical imaging environments (magnetic resonance and computer tomography images) to deal with this challenging real-world problem from the field of forensic medicine.

All the information about this project can be found at http://www.softcomputing.es/socovifi/en/home.php.

4.1.1. Description

One of the main goals of forensic anthropology [Burns07] is to determine the identity of a person from the study of some skeletal remains. In the last few decades, anthropologists have focused their attention on improving those techniques that allow a more accurate identification.

Before making a decision on the identification, it is necessary to follow different processes that let them assign a sex, age, human group, and height to the subject from the study of bones found. Different methodologies have been proposed, according to the features of the different human groups of each region [Iscan05, Aleman97, González07].

Once the sample of candidates for identification is constrained by these preliminary studies, an identification technique is applied. Among them, craniofacial superimposition [Iscan93] is a complex and uncertain forensic process where photographs or video shots of a missing person are compared with the skull that is found. By projecting both photographs on top of each other (or, even better, matching a scanned three-dimensional skull model against the face photo/series of video shots), the forensic anthropologist can try to establish whether that is the same person [Krogman86].

The said process is guided by a number of landmarks located in both the skull and the photograph of the missing person (see Figures 1 and 2). The selected landmarks are located in those parts where the thickness of the soft tissue is low. The goal is to ease their location when the anthropologist must deal with changes in age, weight, and facial expressions.

Figure 1. From left to right, principal facial landmarks: lateral and frontal views.

Figure 2. From left to right, principal craniometric landmarks: lateral and frontal views.

4.1.2. Computer-aided 3D/2D Craniofacial Superimposition Procedure

After one century of development, craniofacial superimposition has become an interdisciplinary research field where computer sciences have acquired a key role as a complement of forensic sciences. Moreover, the availability of new digital equipment (as computers and 3D scanners) has resulted in a significant advance in the applicability of this forensic identification technique [Damas10].

Figure 3. The three stages involved the 3D/2D computer-aided craniofacial superimposition process.

In our view, the whole craniofacial superimposition process is composed of the following three stages (see Figure 3) [Damas09]:

(1) The first stage involves achieving a digital model of the skull and the enhancement of the face image. Obtaining an accurate 3D cranial model has been considered a difficult task by forensic anthropologists in the past. However, it is nowadays an affordable and attainable activity using laser range scanners (Figure 4) like the one used by our team, available in the Physical Anthropology Lab at the University of Granada (Spain). The subject of the identification process, i.e. the skull, is a 3D object. Hence, the use of a skull 3D model instead of a skull 2D image should be preferred because it is definitively a more accurate representation. It has already been shown that 3D models are much more informative in other forensic identification tasks [DeAngelis09]. Concerning the face image, the most recent systems use a 2D digital image. This stage aims to apply image processing techniques [González08] in order to enhance the quality of the image corresponding to the face photograph that was typically provided by the relatives when the person disappeared.

Figure 4. Acquisition of a skull 3D partial view using a Konica-Minolta laser range scanner.

(2) The second stage is the skull-face overlay. It consists of searching for the best overlay of the skull 3D model and the face 2D image achieved during the first stage. The achievement of the right overlay involves two different factors: i) the determination of the real size of the figures (scaling), since it would be impossible to overlay images with a different relative size; and ii) the orientation method for the skull, to make it correspond to the face position in the photograph. There are three possible moves to put that into effect: inclination, extension, and rotation. The overall procedure is usually done by bringing to match some corresponding landmarks on the skull and the face.

(3) Finally, the third stage of the craniofacial superimposition process corresponds to the decision making. Based on the skull-face overlay achieved, the identification decision is made by either judging the matching between the corresponding landmarks in the skull and in the face, or by analyzing the respective profiles. Notice that, the use of computers in this stage aims to support the final identification decision that will be always made by the forensic anthropologist.

4.1.3. Why Should We Use Soft Computing for Craniofacial Superimposition?

In view of the tasks to be performed in the first two craniofacial superimposition stages, it can be seen the relation of the desired procedure with the IR problem in computer vision (see Section 3.3). Besides, from the second and the third stages, we can also draw the underlying uncertainty involved in the whole process. The correspondence between facial and cranial anthropometric landmarks is not always symmetrical and perpendicular (see Figure 5), some landmarks are located in a higher position in the alive person face than in the skull, and some others have not got a directly related landmark in the other set. So, we found a clear partial matching situation and a need for automatic techniques. As a final result, the identification decision can be expressed according to several confidence levels, depending on the chances of the sample (degree of conservation) and of the analytical process put into effect (see Figure 3): “absolute matching”, “absolute mismatching”, “relative matching”, … So, we again find the uncertainty and partial truth involved in the identification process.

Figure 5. From left to right, correspondences between facial and craniometric landmarks: lateral and frontal views.

As seen, different kinds of uncertainty are associated to the current process making the use of FL particularly appealing: the association of the facial and skull anthropometric landmarks is a partial matching process, there is uncertainty on the available knowledge and materials (different degrees of decomposition can affect to the skeleton), partial degrees of truth are present in the resulting final decision, and different information sources must be aggregated to take it.

On the other hand, the whole craniofacial procedure is very time consuming as it is performed by the forensic expert in a iterative trial-and-error way. Besides, there is not a systematic methodology but every expert usually applies his particular knowledge-based process. Hence, there is a strong interest in designing automatic methods to support the forensic anthropologist to put it into effect [Ubelaker00].

In summary, we clearly identify a potential field for the application of SC due to the following reasons:

4.1.4. Our Approach to Soft Computing-based 3D/2D Computer-aided Craniofacial Superimposition

Our approach to deal with the challenge of designing an intelligent system to support the forensic anthropologist in the identification procedure by craniofacial superimposition is based on the use of the following SC techniques in each of the three existing process stages [Damas09, Ibañez09a]:

Stage 1: Face enhancement and skull modeling

In order to accomplish the 3D model of the skull, laser range scanners are equipped with an additional positioning device named rotary table and an appropriate software that permits the 3D reconstruction. Some anthropologists are skilled enough to deal with the set of 3D views and they supervise the procedure of commercial software like RapidFormTM. Nevertheless, these software packages do not always provide the expected outcomes and the anthropologists even have to stitch up manually every couple of adjacent views. Moreover, there are scenarios where it is not even possible to use the rotary table. Hence, a 3D image robust reconstruction method is a real need. However, these is a really complex optimization task, with a huge search space (exhaustive search methods are not useful) that has many local minima (multimodality), but forensic experts demand highly robust and precise results. This complex landscape lead us to propose different evolutionary methods [Santamaría07a,07b,09a], achieving really good results in the automatic alignment of skull range images. A two step pair-wise range IR technique [Bernardini02] was successfully applied to such images. The method includes a pre-alignment stage, that uses a scatter search-based algorithm [Laguna03], and a refinement stage based on the classical iterative closest point algorithm [Besl92]. The method is very robust, indeed it reconstructs the skull 3D model even if there is no turn table and the views are wrongly scanned. An example of a 3D skull model from the Physical Anthropology lab, automatically reconstructed from five partial views by using our evolutionary methods, is shown in Figure 6.

Figure 6. An example of a 3D skull model reconstructed by means of the designed evolutionary range IR methods.

Stage 2: Skull-face overlay

The success of the superimposition technique requires positioning the skull in the same pose of the face as seen in the given photograph. The orientation process is a very challenging and time-consuming part of the craniofacial superimposition technique [Fenton08]. Most of the existing craniofacial superimposition methods are guided by a number of landmarks of the skull and the face. Once these landmarks are available, the skull-face overlay procedure is based on searching for the skull orientation leading to the best matching of the set of landmarks. However, this is again a really complex optimization task, with a highly multimodal landscape (exhaustive search methods are again not useful), and forensic experts again demand highly robust and precise results. This complex landscape lead us to propose different evolutionary methods [Ibañez09a] such as CMA-ES [Hansen01] and different real-coded genetic algorithms [Herrera98], achieving really good results both in performance, competitiveness with human-obtained ones (see overlay results in Figures 7 and 8), and robustness (almost the same overlay results over 30 runs).

Moreover, since in this process the goal is to match two sets of landmarks that belong to two different objects (the face and the skull), there is an inherent uncertainty that must be taken into account. On one hand, the landmark matching uncertainty (not yet modeled in any of our works) will refer to the imprecision involved in the matching of landmarks corresponding to the two different objects, since every pair of landmarks has a different and not fixed matching correspondence. On the other hand, the location uncertainty is related to the extremely difficult task of locating the landmarks [Richtsmeier95] in an invariable place, with the accuracy required for this application. Indeed, every forensic anthropologist is prone to locate the landmarks in a slightly different position. The ambiguity may also arise from reasons like variation in shade distribution depending on light condition during photographing, unsuitable camera focusing, poor image quality, etc. We have proposed the use of fuzzy landmarks [Ibañez09b] to model this kind of uncertainty. This new approach is also relevance to solve the co-planarity problem presented in many overlay cases [Santamaría09b].

Figure 7. From left to right, best skull-face overlay results achieved by the forensic experts and using our automatic evolutionary-based method.

Figure 8. From left to right, best skull-face overlay results achieved by the forensic experts and using our automatic evolutionary-based method.

Overall, the proposed method is fast (it always takes less than 20 seconds) and automatic, and therefore very useful for solving one of the most tedious works (requiring up to 24 hours) performed by the forensic anthropologists. In addition, this method supposed a systematic approach to solve the superimposition problem and in spite of the fact it could need additional improvement, it can already be used in many cases, since it has demonstrated competitive results with the ones achieved by the forensic experts following a manual approach as Figures 7 and 8 show (see [Ibañez09a, Santamaría09b] for some additional results in other real-world identification cases).

Stage 3: Decision Making

Once the skull-face overlay is achieved, the decision making stage can be tackled. The straightforward approach would involve measuring the distances between every pair of landmarks in the face and in the skull. One more time, we have different sources of uncertainty to be tackled in this stage. On the one hand, errors are prone to be accumulated during the process of calibrating the size of the images. On the other hand, we have to propagate the uncertainty of the previous stage and incorporate it in this decision stage. In addition, the final decision will be given together with a confidence degree, resulting in decisions such as: likely positive, likely negative, positive, negative and also undetermined identification. The best way of model this decision making support system is using FL/FSs. We have not modeled this stage yet but we aim to do so in the short future.

4.2. MIBISOC: Medical Imaging Using Bio-inspired and Soft Computing

We also coordinate a Marie Curie International Training Network entitled “MIBISOC: Medical Imaging Using Bio-inspired and Soft Computing” which has been recent granted by the European Commission within the Seventh Framework Program (FP7-PEOPLE-ITN-2008).

The general area of this project deals with the application of intelligent systems constituted by SC-BC techniques to real-world MI applications. Medical imaging (MI) is at the heart of many of today’s improved diagnostic and treatment technologies. Computer-based solutions are vastly more capable of both quantitative measurement of the medical condition and the pre-processing tasks of filtering, sharpening, and focusing image detail. Bio-inspired and Soft Computing (BC, SC) techniques have been successfully applied in each of the fundamental steps of medical image processing and analysis (e.g. restoration, segmentation, registration or tracking). The natural partnership of humans and intelligent systems and machines in MI is to provide the clinician with powerful tools to take better decisions regarding diagnostic and treatment. This project aims to surpass the state of the art approaches applying intelligent systems constituted by SC-BC techniques to real-world MI applications.

The partnership is composed of high quality scientific members, looking for world-wide recognized researchers and high quality technical partners on each area (see Table 1). Direct links between the ITN topics (MI, SC-BC, and SC-BC for MI) and the project partners’ expertise were established, to get together capabilities to face some of the most challenging MI problems by using SC-BC. The network properly balances the presence of research and technical partners, including companies and hospitals, as well as two SMEs, one of them as full network participant.

Table 1. List of the project participants

The network clearly promotes the transverse exchange of knowledge among three different disciplines: medicine, imaging, and computing. Such interdisciplinary approach is show in Figure 9, where the different research areas are linked throughout the MI process as well as the role of every research and technical partner.

Figure 9. Graphical representation of the MI domain subdivided into categories and the assignments of partners and categories according to the partners’ expertise and research plan.

The main goal of the network is to integrate 16 Early Stay Researchers (ESRs) for 36 months in eight leading research groups under the umbrella of a formation program in MI using BC and SC to obtain their PhD degree. The ESRs will learn about a number of important MI problems as well as about the tested and emerging BC and SC techniques, and how to develop methods to solve the former problems by means of the latter techniques as well as to design the associated experiments in a rigorous way. In addition, they will be taught in other complementary skills such as project management, industrial property, etc., by means of the participation in a strictly coordinated international team activity.

The methodology to be followed involves both a theoretical and a practical side. Even though in most cases doctoral studies involve training for research, in this project we would like to focus on training by research. In this way, the outstanding research expertise of the different partners in their respective areas, the practical know-how and the “hands on” scenarios provided by the industrial partners (companies and hospitals), and the experience of all the network participants and associated partners in organizational activities will allow us to implement a high quality training program allowing the exchange of knowledge between the different ESRs selected. The trained ESRs will acquire a strong background for the development of intelligent systems based on BC-SC providing more sophisticated and flexible application-oriented solutions to current MI problems in the clinical and research field. Furthermore, it also aims to provide a transverse research formation from different industrial sectors: scientific research, technology development, practical uses in hospitals, and companies.

With this aim, a personalized, exhaustive and complementary career development plan (PCDP) has been designed for each of the ESRs (see Figure 10), consisting of: i) a personalized research plan based on individual research projects; ii) local and network-wide specific training courses, both in face-to-face and virtual modalities; iii) network’s complementary skills courses, workshops and final conference; and iv) international research stays among the different partners.

Figure 10. Draft template of the described CDP plans

The individual research projects of the ESR are based on novel and attractive research topics of the main research lines of the partners involved and on their collaboration with other research and technical partners. The network will promote the co-supervision of the ESRs’ research training projects and PhD studies. The additional co-supervision from another participant or associated partner will enrich the multidisciplinary and intersectorial aspects of the research carried our by the ESRs and will enhance the collaboration between the network partners. Table 2 shows the individual projects foreseen by the each of the recruited early stage researchers.

The personalized training stage will be started by each of the ESRs once they become contracted in a specific network participant. The core of this first stage will be the academic courses organized locally by her/his hiring partner (mainly its local PhD program), to become trained in that partner’s main topic of expertise. They will be coordinated with two personalized networking activities:

  1. either a first short stay (1-2 months) in other network participant or a secondment (1 month) in a technical associated partner, which in both cases will own a different expertise (e.g., an ESR contracted by an MI partner will do her/his stay either in a SC-BC or SC-BC for MI partner). The ESR will thus complement the concepts acquired by either attending to that other participant’s local courses or benefiting from the associated partner’s technical expertise and “hands-on” scenario; and
  2. an on-line course, available as soon as the ESR joins the network, whose first year will be devoted to introduce her/him in the fundamentals of the two ITN basic disciplines, MI and SC-BC. The syllabus of that first part of the on-line course is coordinated with the local courses and includes all the basic MI and SC-BC concepts required to face the next learning stage and the development of the personalized research training projects, thus properly complementing the skills acquired through the face-to-face modality (i.e., blended learning). Besides, it allows us to establish an active learning modality for the ESR through a virtual community of learners.

Table 2. Individual research projects offered by the host organizations

The latter two activities are very important for the ESR since, from the perspective of a PhD student, working in an existing topic that is of interest to other researchers in the field is positive. Contact with these researchers early in the study can help in guiding her/his formation by focusing on what other researchers deem important. It is also important for the ESRs’ social network, which will become vital for choosing the step following their PhD. Complementary physical and virtual mobility modalities are considered to be implemented through the short secondment and the on-line course.

The network-wide training activities will become the second ESR training stage, mainly starting after the first ESR contract year. They will be coordinated with the basic training themes and the skills taught locally (either at the hiring or the hosting partner) in order the ESR can widen the scope of her/his research interests and knowledge and obtain her/his PhD. The ESRs will: i) enroll in two training courses introducing her/him in different MI application domains solved by SC-BC approaches to get the additional skills needed for his research project; ii) attend to two workshops where (s)he can interact with her/his mates and all the members of the network; iii) enroll in two additional skills courses to complement her/his formation; and iv) attend to a final conference to present her/his research results.

The on-line course will keep on running in this second period for two additional years. During the first of them, the contents taught in the two training courses on MI applications of SC-BC are complemented with a whole syllabus on that topic. The last year is devoted to additional ESR’s active personalized learning through individual and collaborative virtual activities such as new content development resulting from her/his research results, and the maintenance of an Internet portal and of a virtual community of experts on the ITN topic by the ESRs. Finally, we also plan to exchange the hired ESRs among labs for at least a second time in this stage to give them the chance to collaborate with high quality European experts on different MI problems and SC-BC methods helping them to advance in their personalized research training project.

The exhaustive and interdisciplinary proposed training program will provide the European industry with highly qualified researchers able to solve complex MI problems. These researchers will promote new scientific knowledge and technological applications in hospitals, healthcare providers, and technological companies.

For further information about the MIBISOC project, the interested reader is kindly asked to visit: http://www.mibisoc-itn.eu.

5. Concluding Remarks

We have devoted the present manuscript to review some of the research lines developed at the European Centre for Soft Computing. After describing the general aspects of the creation and current structure of the Centre and of its main research and training activities, we have focused on one of the five research units composing it, the AFE unit. Two challenging research projects dealing with the computer vision and medicine fields have been reviewed, showing the potentials and the beneficial characteristics reported by the use of SC to solve different problems in their application domains.

6. References

[Alcalá03]R. Alcalá, J.M. Benítez, J. Casillas, O. Cordón, R. Pérez (2003), Fuzzy Control of HVAC Systems Optimized by Genetic Algorithms. Applied Intelligence 18:2, 155-177
[Aleman97]Alemán, I., Botella, M. C., and Ruíz, L. 1997. Determinación del sexo en el esqueleto postcraneal. Estudio de una población mediterránea actual (in Spanish). Archivo Español de Morfología, 2, 69-79.
[Back97]T. Bäck, D. Fogel, and Z. Michalewicz (Eds.) (1997), Handbook of Evolutionary Computation. Oxford, U.K.: IOP/Oxford Univ. Press.
[Bautista07]J. Bautista, J. Pereira (2007), Ant algorithms for a time and space constrained assembly line balancing problem, European Journal of Operational Research 177, 2016–2032.
[Bernardini02]Bernardini, F. and Rushmeier, H. 2002. The 3D model acquisition pipeline. Computer Graphics Forum 21, 2, 149-172.
[Besl92]Besl, P. and McKay, N. 1992. A method for registration of 3-D shapes. IEEE Transactions on Pattern Analysis and Machine Intelligence 14 (2), 239-256.
[Bonissone97]P.P. Bonissone (1997), Soft computing: the convergence of emerging reasoning technologies, Soft Computing 1 (1), 6-18.
[Burns07]Burns, K. 2007. Forensic Anthropology Training Manual. Prentice-Hall.
[Cas03a]J. Casillas, O. Cordón, F. Herrera, L. Magdalena (Eds.) (2003), Interpretability issues in fuzzy modeling, Vol. 128 of Studies in Fuzziness and Soft Computing, Springer.
[Cas03b]J. Casillas, O. Cordón, F. Herrera and L. Magdalena (Eds.), Accuracy improvements in linguistic fuzzy modeling, Vol. 129 of Studies in Fuzziness and Soft Computing, Springer.
[Chica09a]M. Chica, O. Cordón, S. Damas, J. Bautista (2009), Multiobjective constructive heuristics for the 1/3 variant of the time and space assembly line balancing problem: ACO and randomised greedy, Information Sciences, submitted.
[Chica09b]M. Chica, O. Cordón, S. Damas, J. Bautista (2009), A new diversity induction mechanism for a multi-objective ant colony algorithm to solve a real-world time and space assembly line balancing problema, Memetic Computing, submitted.
[Chica09c]M. Chica, O. Cordón, S. Damas, J. Bautista (2009), Including different kinds of preferences in a multi-objective ant algorithm for time and space assembly line balancing on different Nissan scenarios, Experts Systems with Applications, submitted.
[Cook00]Cook, D. J., Holder, L. B. (2000), Graph-based data mining, IEEE Intelligent Systems 15 (2), 32–41.
[Cordón99]O. Cordón, F. Herrera, L. Sánchez (1999), Solving Electrical Distribution Problems Using Hybrid Evolutionary Data Analysis Techniques. Applied Intelligence 10:1, 5-24
[Cordón01]O. Cordón, F. Herrera, F. Hoffmann, L. Magdalena (2001), Genetic Fuzzy Systems. Evolutionary Tuning and Learning of Fuzzy Knowledge Bases, World Scientific.
[Cordón04]O. Cordón, F. Gomide, F. Herrera, F. Hoffmann, L. Magdalena (2004), Ten Years of Genetic Fuzzy Systems: Current Framework and New Trends, Fuzzy Sets and Systems 141:1, 5-31
[Cordón06a]O. Cordón, S. Damas, J. Santamaría (2006), A Fast and Accurate Approach for 3D Image Registration using the Scatter Search Evolutionary Algorithm, Pattern Recognition Letters 27 (11), 1191–1200.
[Cordón06b]O. Cordón, S. Damas, J. Santamaría (2006), Feature-based image registration by means of the CHC evolutionary algorithm, Image and Vision Computing 22, 525–533.
[Cordón09]L. Sánchez, O. Cordón, A. Quirin, K. Trawinski (2009), Introducing a Genetic Fuzzy Linguistic Combination Method for Bagging Fuzzy Rule-Based Multiclassification Systems, in Fourth IEEE International Workshop on Genetic and Evolving Fuzzy Systems (GEFS), Mieres, submitted.
[Cordón10]O. Cordón, A. Quirin (2010), Comparing Two Genetic Overproduce-and-choose Strategies for Fuzzy Rule-based Multiclassification Systems Generated by Bagging and Mutual Information-based Feature Selection. International Journal of Hybrid and Intelligent Systems, in press.
[Damas09]S. Damas, O. Cordón, O. Ibáñez, J. Santamaría, I. Alemán, M.C. Botella (2009), 3D/2D Computer-aided craniofacial superimposition, International Journal of Legal Medicine, submitted.
[Damas10]S. Damas, O. Cordón, O. Ibáñez, J. Santamaría, I. Alemán, M.C. Botella, F. Navarro (2010), Forensic identification by computer-aided craniofacial superimposition: A survey, ACM Computing Surveys, in press.
[DeAngelis09]De Angelis, D., Sala, R., Cantatore, A., Grandi, M., and Cattaneo, C. 2009. A new computer-assisted technique to aid personal identification. International Journal of Legal Medicine 123, 4 (July), 351-356.
[Dorigo04]M. Dorigo, T. Stützle (2004), Ant Colony Optimization, MIT Press, Cambridge.
[Fenton08]Fenton, T. W., Heard, A. N., and Sauer, N.J. 2008. Skull-photo superimposition and border deaths: identification through exclusion and the failure to exclude. Journal of Forensic Sciences 1, 34-40.
[García07]C. García-Martínez, O. Cordón, F. Herrera (2007), A Taxonomy and an Empirical Análisis of Multiple Objective Ant Colony Optimization Algorithms for Bi-criteria TSP, European Journal of Operational Research 180:1, 116-148.
[Glover03]F. Glover, G.A. Kochenberger (Eds.) (2003), Handbook of Metaheuristics, Kluwer Academic Publishers.
[González07]González-Colmenares, Botella-López, Moreno-Rueda, and Fernández-Cardenete. 2007. Age estimation by a dental method: A comparison of Lamedin's and Prince & Ubelaker's technique. Journal of Forensic Sciences 52 (5), 1156-1160.
[Goshtasby05]A. Goshtasby (2005), 2D and 3D Image Registration, Wiley Interscience.
[Hansen01]N. Hansen, A. Ostermeier (2001), Completely derandomized self-adaptation in evolution strategies, Evolutionary Computation 9 (2), 159–195.
[Herrera98]F. Herrera, M. Lozano, J.L. Verdegay (1998), Tackling real-coded genetic algorithms: operators and tools for the behavioural analysis, Artificial Intelligence Reviews 12 (4), 265–319.
[Holder05]Holder, L. B., Cook, D. J. (2005), Graph-based data mining, In: Wang, J. (Ed.), Encyclopedia of Data Warehousing and Mining, Vol. II. Information Science Reference, Hershey, pp. 943–949.
[Ibañez09a]Ibáñez, O., Cordón, O., Damas, S., Santamaría, J. 2009. An experimental study on the applicability of evolutionary algorithms to craniofacial superimposition in forensic identification. Information Sciences 179, 3998-4028.
[Ibañez09b]Ibáñez, O., Cordón, O., Damas, S., Guadarrama, S., and Santamarìa, J. 2009. A new approach to fuzzy location of cephalometric landmarks in craniofacial superimposition. Proceedings of the IFSA-EUSFLAT World Congress, 195-200, Lisbon, Portugal
[Iscan93]Iscan, M. Y. 1993. Introduction to techniques for photographic comparison. In Forensic Analysis of the Skull, M. Y. Iscan and R. Helmer, Eds. Wiley, New York, USA, 57-90.
[Iscan05]Iscan, M. Y. 2005. Forensic anthropology of sex and body size. Forensic Science International 147, 107-112.
[Karray04]F.O. Karray, C. De Silva (2004), Soft Computing and Intelligent Systems Design: Theory, Tools and Applications, Addison-Wesley.
[Klir96]G.J. Klir, B. Yuan (1996), Fuzzy Sets, Fuzzy Logic, and Fuzzy Systems (Selected Papers by L.A. Zadeh), Advances in Fuzzy Systems-Applications and Theory, vol. 6, World Scientific.
[Koza92]J. Koza (1992), Genetic Programming: On the Programming of Computers by Means of Natural Selection. Cambridge, MA: MIT Press.
[Krogman86]Krogman, W. M., Iscan, M. Y. 1986. The Human Skeleton in Forensic Medicine. Charles C. Thomas, Springfield, USA. 2nd edition.
[Laguna03]Laguna, M. and Martí, R. 2003. Scatter Search: Methodology and Implementations in C. Kluwer Academic Publishers, Boston, USA.
[Magdalena09]L. Magdalena (2009), Soft Computing for Students and for Society, IEEE Computational Intelligence Magazine 4:1, 47-50.
[Pearl88]J. Pearl (1988), Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference, Morgan Kaufmann.
[Quirin08a]A. Quirin, O. Cordón, J. Santamaría, B. Vargas-Quesada, F. Moya-Anegón (2008), A new variant of the Pathfinder algorithm to generate large visual science maps in cubic time, Information Processing and Management 44:4, 1611–1623.
[Quirin08b]A. Quirin, O. Cordón, V.P. Guerrero-Bote, B. Vargas-Quesada, F. Moya-Anegón (2008), A Quick MST-based Algorithm to Obtain Pathfinder Networks(,n-1), Journal of the American Society for Information Science and Technology 59:12, 1912–1924.
[Quirin09a]A. Quirin, O. Cordón (2009), Vmap-Layout, A Layout Algorithm for Drawing Scientograms, In: A. Abraham, A.E. Hassanien and V. Snášel (Eds.), Computational Social Networks Analysis: Trends, Tools and Research Advances. Springer.
[Quirin09b]A. Quirin, O. Cordón, B. Vargas-Quesada, F. Moya-Anegón (2009), Graph-based Data Mining: A New Tool for the Analysis and Comparison of Scientific Domains Represented as Scientograms, Journal of Informetrics, submitted.
[Richtsmeier95]J. Richtsmeier, C. Paik, P. Elfert, T.M. Cole, and F. Dahlman. 1995. Precision, repeatability and validation of the localization of cranial landmarks using computed tomography scans. The Cleft Palate-Craniofacial Journal 32(3), 217–227.
[Romero08]R. Romero Zaliz, C. Rubio, O. Cordón, J.P. Cobb, F. Herrera, I. Zwir (2008), A multi-objective evolutionary conceptual clustering methodology for gene annotation within structural databases: A case of study on the Gene Ontology database. IEEE Transactions on Evolutionary Computation, 12:6, 679-701.
[Rumelhart86]Rumelhart DE, McClelland D (1986) Parallel distributed processing: explorations in the microstructure of cognition. MIT Press, Massachusetts.
[Santamaría07a]Santamaría, J., Cordón, O., Damas, S. 2007. A scatter search-based technique for pair-wise 3D range image registration in forensic anthropology. Soft Computing 11 (9), 819-828.
[Santamaría09a]Santamaría, J., Cordón, O., Damas, S., García-Torres, J. M., and Quirin, A. 2009. Performance evaluation of memetic approaches in 3D reconstruction of forensic objects. Soft Computing 13 (8-9), 893-904.
[Santamaría09b]Santamarìa, J., Cordón, O., Damas, S., Ibáñez, O. 2009. Tackling the coplanarity problem in 3D camera calibration by means of fuzzy landmarks: a performance study in forensic craniofacial superimposition. Proceedings of the IEEE 12th International Conference on Computer Vision, 1686-1693. Kyoto, Japan.
[Schvaneveldt90]Schvaneveldt, R. (Ed.). (1990). Pathfinder associative networks: Studies in knowledge organization. Ablex Publishing Corporation.
[Serrano10]E. Serrano, A. Quirin, J. Botia, O. Cordón (2010), Debugging complex software systems by means of pathfinder networks, Information Sciences, in press.
[Trawinski09]K. Trawinski, O. Cordón, A. Quirin (2009), On the combination of accuracy and diversity measures for genetic selection of bagging fuzzy rule-based multiclassification systems, in Nineth International Conference on Intelligent Systems Design and Applications (ISDA), Pisa, 2009.
[Ubelaker00]Ubelaker, D. H. 2000. A history of Smithsonian-FBI collaboration in forensic anthropology, especially in regard to facial imagery. Forensic Science Communications 2:4, [online].
[Vargas07]Vargas-Quesada, B., & Moya-Anegón, F. (2007). Visualizing the structure of science. New York: Springer.
[Washio03]Washio, T., Motoda, H. (2003). State of the art of graph-based data mining, SIGKDD Explorations 5 (1), 59–68.
[Zadeh94]Zadeh, L.A. (1994), Fuzzy Logic, Neural Networks, and Soft Computing, Communications of the ACM 37(3), 77-84.

About the Author

Oscar Cordón has been for 15 years an internationally recognized contributor to R&D Programs in soft computing (especially in fuzzy systems, evolutionary algorithms, and ant colony optimization and other metaheuristics). He received his Ph.D. (1997) in Computer Science from the University of Granada, Spain, where he has been an assistant professor since 1995 and associate professor since 2001. He was the founder and leader of the Virtual Learning Center (CEVUG) between 2001 and 2005, and was awarded with the Young Researcher Career Award in 2004. Since April 2006, he is the Principal Researcher of the "Applications of Fuzzy Logic and Evolutionary Algorithms" research unit at the European Centre for Soft Computing, located in Mieres, Spain. In December, 2008, he received the Full Professor accreditation from the Spanish Quality Evaluation Agency.

He has published 220 peer-reviewed scientific publications, including a co-authored book on genetic fuzzy systems and 48 SCI-JCR-indexed journal papers. By December 14th 2009, his publications had received 1117 citations, he had an h-index of 18, and he was included in the 1% of the most cited researchers in the world (source: Thomson’s Web of Knowledge; h-index=28 in Google Scholar). He has participated in 29 national and European research projects and contracts, having coordinated 16 of them. Dr. Cordón is an associate editor of two SCI-JCR-indexed journals (IEEE TFS and IJAR) and of a new one launched by Willey (WIRE DMKD), as well as editorial board member of four other journals. He has supervised 11 PhD dissertations. He has also edited 7 special issues of international journals and 3 research books. He has organized 4 international workshops and conferences, and 10 special sessions in conferences.

He is member of the Fuzzy Systems TC, IEEE Computation Intelligence Society (CIS) since 2004. In that year, he created the Genetic Fuzzy Systems TF and chaired it until 2007. He has been the general co-chair of two of the four International Workshops organized by that TF (GFS2005 and GEFS2010), all of them technically co-sponsored by the IEEE CIS. He is also a member of the Future Directions in Fuzzy Sets and Systems TF, the Standards Technical Committee, and the Graduate Student Research Grants sub-committee. An IEEE TFS Associate Editor (AE) since 2008, he was recognized as Outstanding AE during his first year of service. He served as Publicity co-chair of the IEEE SCCI 2009 and as Special Sessions co-chair of the CEC for WCCI2010. He has recently been elected as IEEE CIS AdCom member for the period 2010-2012. Besides, he was member of the Eusflat board (Treasurer) from 2005 to 2007 and he is the Calls Manager in the new Board (2009-2011). He has been Finance co-chair of the 2009 IFSA-Eusflat Joint Conference and Area Chair for Evolutionary Algorithms in IPMU2010. Moreover, he is a member of the IPC of more than 90 conferences and frequently acts as a reviewer for more than 25 international journals.

His current main research interests are in the fields of soft computing for forensic anthropology and medical imaging; genetic fuzzy systems; soft computing and visual science maps; and evolutionary computation, ant colony optimization and other metaheuristics.

Sergio Damas received his Ph.D. (2003) in Computer Science from the University of Granada, Spain, where he has been an assistant professor since 1995 and permanent professor since 2007. His Ph.D. dissertation dealt with the applications of evolutionary algorithms and other metaheuristics to the image registration problem. Since April 2006, he is Associate Researcher at the recently created European Centre for Soft Computing (Mieres, Spain). In January 2008, he received the Associate Professor accreditation from the Spanish Quality Evaluation Agency.

He has published more than 50 peer-reviewed scientific publications, including 9 SCI-JCR-indexed journal papers in prestigious journals such as ACM Computing Surveys (the one with the highest impact factor among the Computer Science journals) , Information Sciences, Pattern Recognition Letters, Image and Vision Computing, and Soft Computing. He has edited a special issue in the International Journal of Approximate Reasoning (IJAR) on “Soft Computing in Image Processing”, also indexed in the SCI-JCR. He has organized different international workshops and special sessions in national and international conferences on soft computing and computer vision. He has supervised a PhD dissertation on the latter topic. He has participated in 10 national and European research projects and contracts. Moreover, he is a member of the IPC of more than 10 conferences and frequently acts as a reviewer for more than 10 international journals in the fields of soft computing and computer vision, as IEEE Transactions on Image Processing, Evolutionary Computation, Fuzzy Sets and Systems, Applied Soft Computing, etc.

Raúl del Coso is an expert in technological innovation with experience in multidisciplinary research and project management, and with expertise in the commercialization of new technologies and business development. Since March 2007, he is Operations Manager at the recently created European Centre for Soft Computing (Mieres, Spain).

He received a Ms.C (1998) in Theoretical Physics from the Universidad Autónoma de Madrid and he carried out his PhD (2004) in Applied Phyiscs in the Instituto de Óptica of the CSIC. In December 2006, he graduated from Imperial College Business School's part-time Executive MBA with Distinction. Raul has experience in nanomaterials, photonics, medical technologies and in the biopharmaceutical sector. He has made research visits to Universität Konstanz in Germany (7 months), to Southampton University (U.K.) and to Université Paul-Sabatier in Toulouse (France). In 2004 he worked as manager of innovation projects for SMEs and regional organizations. He has worked as Project and Technology Transfer Manager for more than two years for a group of multidisciplinary projects at Imperial College London. He has participated in more than ten R&D projects with public and private funding from European, national and regional sources.

Óscar Ibáñez Panizo was born in Ponferrada (León) in 1981. He received his Technical Engineering degree (2002) in Computer Science at the Pontificia University of Salamanca and the M.Sc. degree (2006) in Computer Science at the University of La Coruña.

He was a member of the VARPA and RNASA/IMEDIR research groups (University of La Coruña, Spain) from 2004-2006 and 2006-2008, respectively. Since March 2008, he is a research assistant in the Applications of Fuzzy Logic and Evolutionary Algorithms Unit at the European Centre for Soft Computing (Mieres, Spain).

He has published more than 20 papers between international journals, peer-reviewed conferences, and international books. His main interests are focus on Evolutionary Algorithms, Artificial Neural Networks, Fuzzy Logic and their applications in image analysis.

Carmen Peña received her degree (2002) and her Ph.D. (2007) both in Organic Chemistry from the University of Oviedo. Dr. Peña's research was focused on the field of supramolecular chemistry, specifically on the synthesis of quiral polyamine receptors and on the study of their properties in enantioselective complexation of biologically interesting anions, such as DNA', among others. She has published 11 contributions in citation index journals and conferences, and has also a patent (P-200802138). She was recently awarded with the San Alberto's Prize 2008, of the Chemists Association of Asturias, for her research work: "Synthesis and Stereoselective DNA Binding Abilities of New Optically Active Open-Chain Polyamines". In April 2007, she joined the company "Entrechem Biotechnology", spin-off of the University of Oviedo, where she worked in new drug synthesis and discovery. In January 2008 she started working in the consulting company "Impulso Industrial Alternativo" where she managed different R&D projects for companies of different activity sectors such as: CEPSA, Vitro CristalGlass, Lider IT Consulting, Aleastur, Verdifresh, STE Pharma Systems, etc. Since July 2008 she is Project Manager at the European Centre for Soft Computing (Mieres, Spain).