Decision Making and Soft Computing

The following sections are included:

• FOREWORD
• CONTENTS

We try to provide a tentative assessment of the role of fuzzy sets in decision analysis. We discuss membership functions, aggregation operations, linguistic variables, fuzzy intervals and valued preference relations. The importance of the notion of bipolarity and the potential of qualitative evaluation methods are also pointed out. We take a critical standpoint on the state of the art, in order to highlight the actual achievements and point out research directions for the future.

In numerous real-world problems including a broad range of decision-making tasks, we are faced with a diversity of locally available distributed sources of data and expert knowledge, with which one has to interact, reconcile and form a global and user-oriented model of the system under consideration. While the technology of Soft Computing has been playing a vital and highly visible role with this regard, there are still a number of challenges inherently manifesting in these problems when dealing with collaboration, reconciliation, and efficient fusion of various sources of knowledge. To prudently address these problems, in this study, we introduce a concept of granular fuzzy systems forming an essential generalization of fuzzy systems pursued in Soft Computing. Information granularity of fuzzy sets used in these models is formalized in the framework of Granular Computing. We briefly elaborate on the fundamentals of Granular Computing including (i) a principle of justifiable granularity, (ii) an allocation of information granularity being sought as an essential design asset, and (iii) an emergence of higher type and higher order information granules in investigations of hierarchical architectures of systems. We show the roles of these principles in the analysis and synthesis of granular fuzzy systems. A class of group decision-making problems is studied in detail. We investigate granular AHP models and demonstrate a pivotal role of information granularity in the generalization of these constructs.

Note from Publisher: This chapter consists of abstract only.

The decision support system is a software system that aims to facilitate and/or enhance the process of decision making activities at any organizational level. Decision support system has moved from the traditional role of supporting traditional business process into supporting intelligent decision making. Most of the proposed software architectures for decision support system are domain-specific software architecture. We have proposed simplified software architecture for decision support systems that is suitable as a generic software architectural style for decision support systems. We believe that the proposed software architectural style can work for many different domain problems for decision support systems.

Early debt collection systems aim at collecting payments from the creditors with a minimum cost before the legal procedure. Our study develops a fuzzy inference system for early debt collection problems including the inputs amount of loan, wealth of debtor, past history of debtor, and amount of other debts and the output possibility of repaying the debt and the way of communication. Thus maximum collection of debts before legal process with minimum cots can be achieved.

This paper presents a new architecture for a Spatio-Temporal Decision Support System. It was designed taking into account epidemiological aspects for decision making in public health management. The main goal is analyze the spatial and spatio-temporal features of a geographic area and make decisions about prevention and control of a disease dissemination on that area. An example of application of this architecture is presented to construct a system for real aids data is presented, as well as the results obtained from it.

In this study, supplier reliability is considered as a criterion for selecting the suppliers. Although, there are dozens of studies in the literature that use different models on supplier performance evaluation, only in a few of these studies supplier's reliability is taken as a parameter of the evaluation. The goal of this study is to determine the factors which play important role on the reliability of suppliers and analyze the importance degrees of these factors. For this purpose, fuzzy DEMATEL method which is frequently preferred in the literature is used in order to analyze interaction among the factors.

We present a new possibilistic mean-variance model using the Fuzzy Laplace distribution (PMVFL). We generated a sequence of results and concluded that results showed an expected behavior of model of possibilistic mean-variance. When we increase the VaR (Value at Risk), in other words, when we consider further loss of market value, we mean that the risk rate will be higher, i.e., larger return rate, higher will be risk rate, this fact has been demonstrated in model.

Hesitant fuzzy sets have been developed to handle the situations where a set of values are possible in the definition process of the membership of an element. Hesitant fuzzy linguistic term sets provide a linguistic and computational basis to increase the richness of linguistic elicitation based on the fuzzy linguistic approach and the use of context-free grammars by using comparative terms. In this paper, we propose a multicriteria method based on hesitant fuzzy linguistic term sets and apply it to a supplier selection problem.

To weaken the effects of those experts’ weights that far away from the center of opinions in decision-making approach for assessing the performances of candidates in many major competitions, this paper proposes a weight determination method of experts in group decision based on the difference value, which can effectively weaken the effects of extreme scores in group decision making and centralize the experts’ opinions for achieving consensus. A case study is demonstrated to illustrate this method and the results indicate the effectiveness.

The evaluation of medical imaging devices is a critical issue for both biomedical engineers and health-care investors. This study proposes a new technique to assess common medical imaging devices using type-2 fuzzy multi-criteria decision making approach. The evaluation criteria were characterized by the interviews with the experts. A Gaussian type-2 Fuzzy membership function was assigned for each interval of the evaluation. TOPSIS algorithm was applied to our system using type-2 Fuzzy numbers. The results were classified with the Wu and Mendel's ranking method. The ranking of device alternatives highlighted the accurate order of future imaging technologies with the fuzzy behavior of medical investments in conjunction with the requirements of the clinicians and the engineers.

In the knowledge economy, a key source of sustainable competitive advantage relies on the way to create, share, and utilize knowledge. This paper presents an application of the interval type-2 TOPSIS method used to select the most appropriate tool to support knowledge management (KM) activities in a healthcare system. The method provides us with a useful way to handle fuzzy multiple attributes group decision-making problems in a more flexible and more intelligent manner to analyze and compare KM tools in the software market. A case study is given to demonstrate the potential of the methodology.

Today, most companies monitor the satisfaction of their employees. Companies are trying to measure the satisfaction of their employees and with this aspect they are making efforts to increase workload and work quality, which is beneficial for the company itself. This article proposes a fuzzy tool for assessing the employee's satisfaction within the company. A part of the fuzzy expert system is a tool for evaluating satisfaction questionnaires. All the parts of the system are introduced and verified on a specific example.

Every company occasionally deals with the problem of hiring new staff. The issue of selecting of a suitable candidate for a job is often very complicated and many criteria enter the selection and hiring process. This paper proposes an expert system for the selection of the most suitable applicants for employment. A part of the proposed approach is the definition of criteria for selection of suitable applicants and lining to the database of all applicants. Furthermore, this paper introduces each part of the proposed expert system and all the steps are verified on a specific example.

Environmental factors are the difficulties of entrepreneurs when facing in the whole process of entrepreneurship. Due to the dynamic complexity of environmental factors and the limitations of entrepreneurial recognition, an entrepreneur will find it difficult to comprehend the business environment. The analysis of entrepreneurial environment assessment is born based on the necessity, effectiveness of the proposed model, giving a pioneering project environmental assessment, establishing venture project dynamic environmental evaluation algorithm, which application will be approved by examples.

Risk assessment is one of the most important skills that pilots are expected to acquire to ensure the safe and successful management of flight. The traditional approach to the development of these skills requires pilots to directly engage with potentially hazardous events. This paper develops dual adaptive neuro-fuzzy inference system (D-ANFIS) in civil aviation as context for pilot's risk assessment, which can help reduce the risk to the aircraft aviation flight, while maintaining operational performance. It was concluded that exposure to hazards within a simulated environment could provide the basis for the development of risk assessment skills amongst less experienced pilots.

Aiming to handle logical formulae with equality in lattice-valued logic, this paper focuses on α-paramodulation for lattice-valued logic with equality. Firstly, the axioms of equality and their related properties in lattice-valued logic are presented. Then the concept of α-paramodulation is given, which is the essential rule for processing logical formulae with equality. Finally, the soundness of α-paramodulation is established.

Based on 6-elements linguistic truth-valued lattice implication algebras this paper discusses 6-elements linguistic truth-valued first-order logic system. We give some equivalent formal of 6-elements linguistic truth-valued first-order logic system is given. We provide 6-elements linguistic truth-valued first-order logic of general implications formula and the law of negative transformation.

In the present work, we assess the application of the Nash-Particle Swarm Optimization (Nash-PSO) algorithm to the analysis of timber markets in the Amazon forest within a game theoretical framework. The usage of the PSO algorithm and the game theory's best response concept are the bases of the Nash-PSO algorithm, implemented for such analysis. With the Nash-PSO algorithm it is possible to analyze the interactions of players in a continuous space of strategies, for non-linear objective functions with a fast and accurate convergence. The results also demonstrate the viability of the Nash-PSO algorithm in the estimation of real values for government investment in forest areas.

In this paper, three risk indicators on road safety (i.e., traffic fatalities per vehicle, per traveled kilometer and per inhabitants) are combined into a composite indicator in order to assess the overall fatality risk for the 27 Brazilian states. The so-called multiple layer data envelopment analysis (DEA) model is used in this respect. Given their remarkable diversity in terms of road safety context, the states are first clustered (into four clusters) and next, a range of bootstrapped scores is generated to manifest the estimated variability in the road safety performance of each state. Bootstrapping the original DEA scores showed to be a useful strategy to assess the robustness of the states’ ranking, which is subjected to uncertainties in input information and DEA model determinism.

In the article we alternatively develop forecasting models based on the Box-Jenkins methodology and on the neural approach based on classic and fuzzy logic radial basis function neural networks. We evaluate statistical and neuronal forecasting models for monthly platinum price time series data. In the direct comparison between statistical and neural models, the experiment shows that the neural approach clearly improve the forecast accuracy. Following fruitful applications of neural networks to predict financial data this work goes on. Both approaches are merged into one output to predict the final forecast values. The proposed novel approach deals with nonlinear estimate of various radial basis function neural networks.

This paper presents a feature selection method based on genetic algorithm for unbalanced data. This method improves the fitness function through using the evaluation criterion G-mean for unbalanced data instead of total classification accuracy in order to improve the recognition rate of the minor class. Experimental results on several UCI datasets show that the performance of the proposed method outperforms classic genetic algorithm. It not only reduces the feature dimension effectively, but also improves the precision of the minor class.

The resources and possibilities to reproduce objective information through infography methods are numberless. This article presents interactive infography contributions to represent data as a result of a systematic mapping in the area of online social networks analysis. We tried to evidence how interactive infography resources can synthetize complex information in a simple visual representation.

This paper introduces preliminary concepts on interval fuzzy Bayesian games, based on interval-valued fuzzy probabilities for modeling of types of agents involved in the interaction. The interval-valued fuzzy probabilities are given by symmetric triangular interval fuzzy numbers, inspired by Buckley and Eslami, who considered an arithmetic restriction on the interval [0, 1] in their representation of uncertain probabilities by fuzzy numbers.

Rough set is an useful tool to characterize uncertain information. It has been widely applied to feature selection. In this paper, we extend the information entropy into the kernel fuzzy rough set. Then, we construct the fast feature selection algorithm based on conditional entropy. Experimental results show that the conditional entropy-based feature selection algorithm can improve the classification performance with few features in most of the cases.

We present an idea to group time series according to the course of their local trends that can be well captured by the F¹-transform. On the basis of an adjoint time series consisting of a sequence of F¹-transform components, we form a grouping of time series with closely related trends. This enables us to forecast trend of one selected principal time series and on the basis of it, to forecast trends of the other time series from this grouping. This is realized using the methods of fuzzy natural logic, namely automatic generation of linguistic description from the data and then deriving a conclusion using the perception based logical deduction.

We address the use of two strategies for edge detection in polarimetric Synthetic Aperture Radar – SAR imagery. Both approaches stem from the realm of optical images: Canny and one inspired by the Law of Universal Gravitation. Images are filtered prior to edge detection by two procedures: Lee and Torres filters. Two types of neighborhoods are employed.

In this paper we proposed new estimators of parameters for a Naive Bayes Classifier based on Beta Distributions. Equations were obtained for these estimators using an EM-like algorithm and they provide numerical estimates for those parameters. Furthermore, two forms for that Naive Bayes Classifier were presented.

Evolving Fuzzy Neural Networks (EFuNNs) are dynamic connectionist feed forward networks. Several paper can be found in the literature in which EFuNN reach better results than other methods. However, only one paper was found in which EFuNN results were analyzed with respect to some statistical distributions of data. This study has as goal to complement the previous study, evaluating the EFuNN performance using four other statistical distributions. Results of assessment are provided and show different accuracy according to the statistical distribution of data.

Assessment systems of training based on Virtual Reality are used to measure users' skills when performing a procedure. This kind of systems can demand expressive time of CPU. A solution was pointed out using an architecture for assessment based on embedded systems. The goal of this paper is to analyze the efficiency of hardware architecture in producing good results, using two previously proposed online methods based on fuzzy sets.

There has been an increasing amount of research in the relationship between environmental factors and fishing yield. This paper adds to the body of knowledge by developing a new model for forecasting fishing yield. The model combines fishery domain expert knowledge, marine environmental factor data such as water temperature, chlorophyll concentration and sea surface level as base data and applies cluster analysis that incorporates function fitting and nonlinear regression for data analysis and processing. The model is tested for forecast accuracy and the test result is compared with those using RBF and SVM, the two methods commonly used for similar purposes. The comparison result reveals this new model increases both the accuracy in fishery forecast and the reliability in guiding fishery production and related activities. It can also help explore and discover the distribution of fishing grounds.

In this paper, an efficient online learning approach is proposed for Support Vector Regression (SVR) by combining Feature Vector Selection (FVS) and incremental learning. FVS is used to reduce the size of the training data set and serves as model update criterion. Incremental learning can “adiabatically” add a new Feature Vector (FV) in the model, while retaining the Kuhn-Tucker conditions. The proposed approach can be applied for both online training & learning and offline training & online learning. The results on a real case study concerning data for anomaly prediction in a component of a power generation system show the satisfactory performance and efficiency of this learning paradigm.

Based on the academic ideas of resolution-based automated reasoning and the previously established research work on general form of α-resolution based automated reasoning schemes in the framework of lattice-valued logic with truth-values in a lattice algebraic structure-lattice implication algebras (LIA), this paper is focused on investigating α-n(t)-ary resolution dynamic automated reasoning method in lattice-valued propositional logic LP(X) based in LIA. In this paper, the definition of α-ordered linear minimal resolution in LP(X) is introduced firstly. Then, its soundness and completeness are proved. One of key issues for α-n(t)-ary resolution dynamic automated reasoning for LP(X) is how to choose generalized literals. It will guide how to choose generalized literals to improve the resolution efficiency in LP(X). This lays the foundation for the further study on α-n(t)-ary resolution automated reasoning.

As a continuation and extension of the established work on binary resolution at certain truth-value level (called α-resolution), this paper introduces non-clausal multi-ary α-generalized resolution principle and deduction for lattice-valued propositional logic LP(X) based on lattice implication algebra, which is essentially a non-clausal generalized resolution avoiding the reduction to normal clausal form. Non-clausal multi-ary α-generalized resolution deduction in LP(X) is then proved to be sound and complete.

The aim of this contribution is to discuss the characterizations of ℒ-semilinear spaces which are generated by strong linearly independent vectors. We will show that the basis in ℒ-semilinear spaces which are generated by strong linearly independent vectors is also strong linearly independent.

This paper mainly focus on building the prime ideals of non regular residuated lattices. Firstly, two types prime ideals of a residuated lattice are introduced, the relations between the two types ideals are studied, in some special residuated lattices (such as MTL-algebras, lattice implication algebras, BL-algebras), prime ideal and prime ideal of the second kind are coincide. Secondly, the notions of fuzzy prime ideal and fuzzy prime ideal of the second kind on a residuated lattice are introduced, aiming at the relation between prime ideal and prime ideal of the second kind, we mainly investigated the fuzzy prime ideal of the second kind.

In 2013, Bergamaschi and Santiago proposed Strongly Prime Fuzzy(SP) ideals for commutative and noncommutative rings with unity, and investigated their properties. This paper goes a step further since it provides the concept of Strongly Prime Radical of a fuzzy ideal and its properties are investigated. It is shown that Zadeh's extension preserves strongly prime radicals. Also, a version of Theorem of Correspondence for strongly prime fuzzy ideals is proved.

The theory of filters and fuzzy filters in logical algebras play a vital role in reasoning mechanism in information sciences, computer sciences, theory of control, artificial intelligence and many other important fields. We introduce the concept of fuzzy sub positive implicative filters of residuated lattice and investigate the properties of it, and further characterize the fuzzy sub positive implicative filters by proposing the equivalent conditions that a fuzzy filters to be a fuzzy sub positive implicative filters.

Femtocells has been regarded as one of the most promising approaches to improve indoor coverage and network capacity. The method of decentralized spectrum allocation between users has become more efficient. An adaptive spectrum allocation scheme based on matrix game is proposed, in which femto and macrocell base stations are players and the same spectrum is the resource players will choose to assign users, to minimize the affected interference among each other. The equilibrium is the output of the game, which is also the optimal spectrum allocation manner. The sub-optimal solution is also given to avoid that the equilibrium may not exist. And the comparison results show that the proposed scheme might be a solution for efficiently allocating the spectrum in hierarchical cell networks, as the improvement in terms of throughput, outage probability had been achieved.

Linguistic truth-valued intuitionistic fuzzy lattice ℒℐ_2n = (LI_2n, ∪, ∩, →, ((h_n, t), (h_n, f))), ((h₁, t), (h₁, f)) based on linguistic truth-valued lattice implication algebra has some special properties. We proof that linguistic truth-valued intuitionistic fuzzy lattice is a triangle algebra. We obtain some triangle algebra structure of ℒℐ_2n.

We investigate the Cauchy problem for an ordinary differential equation (ODE). We propose two new schemes that compute an approximate solution. Both of them are based on the technique of the second degree F-transform. The quality of new schemes is compared with the exact solution and the Euler method.

This paper introduces the notion of interval migrative functions. Also, we show a necessary and sufficient condition to a interval function to be migrative and that the interval canonical representation of a migrative function f (in the usual sense) is an interval migrative function and preserves some properties of f.

There is controversy regarding the use of the similarity functions proposed in the literature to compare generalized trapezoidal fuzzy numbers since conflicting similarity values are sometimes output for the same pair of fuzzy numbers. In this paper we propose a similarity function aimed at establishing a consensus. It accounts for the different approaches of all the similarity functions. It also has better properties and can easily incorporate new parameters for future improvements. The analysis is carried out on the basis of a large and representative set of pairs of trapezoidal fuzzy numbers.

This paper aims at establishing an outline of 2ⁿ–valued propositional calculus (2ⁿP) to set logical foundation for big data science. After introducing the characteristics of big data and the significance of researching on big data, we briefly analyze features of famous L_n system. This paper specifies 2ⁿP from logical semantic and syntax. We firstly define connectives including negation ¬ and disjunction ∨, and define conjunction ∧ and implications → based on them; we prove the {¬, →} is adequate set of connective. Then we structure the axiom set including all axioms of classical logic, and prove modus ponens and the consistent of 2ⁿP, yet give the soundness theorem and the adequate theorem of 2ⁿP.

This paper presents the graphical illustration of the Boolean consistent real-valued relations on the example of two two-dimensional objects. Consistent real-valued relations are based on the real-valued realization of the Boolean algebra.

This paper presents an α-resolution method for lattice-valued horn generalized clauses in lattice-valued propositional logic system ℒP(X) based on lattice implication algebra. In this approach, We give lattice-valued horn generalized clause and the correlative concepts in ℒP(X). The α-resolution of two lattice-valued horn generalized clauses is also represented in ℒP(X). It reflects the resolution rules in a resolution process, which aims at deleting α-resolution literals and obtaining a resolvent. This method can provide an efficient tool for automated reasoning in lattice-valued propositional logic system and lattice-valued first-order logic system.

Boolean networks are models of complex dynamical systems. Modelling complex systems with Boolean networks is adequate in situations in which binary view is valid. If the situation is not “black and white”, Boolean networks are inadequate/oversimplified models. For modelling real world complex systems, Boolean networks lack descriptive power – fuzzification of the model is required. Boolean networks are fuzzified using the interpolative Boolean algebra. Fuzzy model keeps the Boolean frame.

Data clustering is widely used in management decision and other fields. The traditional methods of data clustering based on offline calculating are faced with the challenge of speed and cost in the process of emergency decision making of social problem. Therefore, it's necessary to research a new method of data clustering as online using intelligent technologies. On discussing on the problem of data clustering, analyzing the basic assumption and selection criteria, a dynamic program of data clustering based online is put forward in this paper. The theory of dynamic learning is used in establishing an algorithm of data clustering.

The aggregating fuzzy (S,N)-subimplications is obtained by the OWA-operator performed over the families of triangular sub(co)norms along with fuzzy negations. (S,N)-subimplications are characterized by the generalized associativity and distributive properties together with extensions of the exchange and neutrality principles. As the main results, these families of subimplications extend related S-implications by preserving their corresponding properties. We also discuss the action of automorphisms on such fuzzy implication classes.

In this paper, we studied the classes of quasi-homogeneous overlap functions. In which we demonstrate that all the class of quasi-homogeneous overlap functions properly contains the class of quasi-homogeneous t-norms, that is, all quasi-homogeneous t-norm is a quasi-homogeneous overlap function, but not all overlap function quasi-homogeneous is a quasi-homogeneous t-norm.

Within this contribution we establish a theoretical background necessary for studying inverse limits of fuzzy dynamical systems induced by crisp (non-fuzzy) ones. First, we elaborate topological properties of the space of fuzzy sets, such as connectedness and compactness. Second, we focus our attention on dynamical conditions sufficient for the existence of indecomposable continua in the inverse limit space.

In this paper we apply a method to extend n-dimensional lattice-valued fuzzy negations by preserving the largest possible number of properties of these negations which are invariants under homomorphisms. Further, we also prove some related results and properties.

In this paper we consider the notion of Fuzzy Lattices which was introduced by Chon in 2009. We define the fuzzy homomorphism between fuzzy lattices and proved some results of fuzzy homomorphism on bounded fuzzy lattices. Also, we prove some results involving fuzzy homomorphism and ideals.

In this paper, some results about interval uninorms with the additional property of monotonicity inclusion are introduced; e.g construction of interval uninorms from usual uninorms and constructions of interval t-norms and t-conorms from interval uninorms. It is also shown that the neutral element of this type of interval uninorm must be a degenerate interval.

New aggregation operators are introduced by using distance measures with the ordered weighted average, the probability and the weighted average. This approach is developed by using the Minkowski distance which uses generalized means in the aggregation process. In order to do this, it is introduced the generalized probabilistic ordered weighted averaging weighted averaging distance (GPOWAWAD) operator. It provides a parameterized family of aggregation operators between the minimum distance and the maximum one considering subjective and objective information in the analysis. Some of its main properties and particular cases are also studied.

In this paper, a rule base representation with certainty factors is proposed firstly along with its inference method. Such a rule base is designed with certainty factors embedded in the consequence terms, rule terms as well as in the antecedent terms, which is shown to be capable of capturing uncertainty. As the evidential reasoning approach is applied to the rule combination, the overall representation and inference framework can be applied in rule based system for human decision making due to the fact. A numerical example is examined to show the implementation process of the proposed method, as comparing with a classical approache we can see its high perfprmance.

The polarity of concepts and the dialectic process by which its meaning emerges has been subject of interest since the ancient Greeks. Recently, the term Bipolarity has been used in social and mathematical sciences, referring to the measurement of the meaning of concepts. It is claimed that the measuring process has to consider at least an associated pair of meaningful opposites, such that some type of structure is used to analyze the aspect of reality that is being modeled. From this point of view, we take a quick overview on the genealogy of Bipolarity, discussing some ideas about the nature of negative knowledge, and how it has been examined recently, and not so recently, by the mathematical community.

On the basis of multiary α-resolution principle, a multiary α-resolution automated reasoning method--α-semi-lock semantic resolution method is studied in lattice-valued propositional logic system LP(X) based on lattice implication algebra. Concretely, α-semi-lock semantic resolution method is established in LP(X), as well as its soundness and condition completeness.

This article deals with propositional fuzzy modal logic with evaluated syntax based on MV-algebras. We focus on its semantic theory from the viewpoint of Pavelka's graded semantics of propositional fuzzy logic, investigate the L-tautologies based on different Kripke frames. We also define the notion of L-semantic consequence operation, its some basic properties are presented.

This paper proposes a formal framework which represents the composite human activity under consideration by a hierarchical ordering structure and discusses how they can be modelled and transferred into a formal syntactical logical formula, i.e., logical predicate algebra. This has placed a foundation for recognizing the composite activity based on the transformed logical formulas using automated reasoning methodology.

The In-Core Fuel Management Optimization (ICFMO) is a prominent and real-world combinatorial problem in Nuclear Engineering, with a large number of sub-optimal solutions, disconnected feasible regions and approximation hazards. For the sake of previous validation, optimization techniques are applied to benchmark combinatorial problems prior to applications to the ICFMO itself. In the present work, the investigation on the application of Particle Swarm Optimization (PSO) to the Quadratic Assignment Problem (QAP) is reported. The Random Keys (RK), an encoding model used to map particles' positions in a continuous search space into combinatorial solutions, allowed promising results without constructive heuristics and local search. The application of PSO with RK to the QAP forms a basis for further investigation on the RK encoding scheme for the ICFMO.

The current communication technologies have resulted in a modern world characterized by a remarkable increase in social interactions. In this new context and because of the globalization of all human activities make the collective participation in decision-making processes take an increasingly prominent role. In this paper, a new method for group decision making from a set of imprecise opinions called “moviQuest Decision Making” (MQDM), is presented. This method allows to integrate the opinions of heterogeneous groups of agents for iterative collective decision making.

In linguistic decision making problems, the set of alternatives are assessed by means of linguistic terms, implying processes of Computing with Words (CWW). The 2-tuple linguistic model provides a computational model that offers linguistic results in the original linguistic domain in a precise way. Furthermore, this model has been extended to carry out processes of CWW in complex decision frameworks. Despite these advantages, this model and its extensions have not been developed in a software tool suite to facilitate the resolution of linguistic decision making problems. In this contribution, we present FLINTSTONES, a fuzzy linguistic decision tools enhancement suite to solve linguistic decision making problems based on the 2-tuple linguistic model and its extensions as well as the FLINTSTONES website.

α-resolution principle based on lattice-valued logic is focused on in this paper, which is a great extension of resolution principle based on classical logic. Concretely, based on α-Group resolution principle in lattice-valued logic, α-Group lock resolution principle is introduced, aiming to improve the efficiency of resolution and the ability of resolution. Then both the soundness and the completeness theorems of α-Group lock resolution principle are proved in lattice-valued propositional logic LP(X). The example given in lattice-valued propositional logic LP(X) illustrates that α-Group lock resolution method is, to some extent, more resolution efficiency than α-Group resolution method and more resolution ability than α-lock resolution method, which only permits binary resolution in lattice-valued logic. This work will provide theoretical foundations for the establishment of automated reasoning algorithm and the further applications of automated reasoning and decision making under uncertainty.

The work considers the problem of solving a system of fuzzy relational equations and introduces the concepts of characteristic matrix and attainable variables. It is proved that maximal solutions of the system correspond to irredundant coverings of characteristic matrix.

To find the influence on clustering result of load curves using different clustering algorithms with different data normalization methods, seven data normalization methods are used with k-means, fuzzy c-means, SOM clustering algorithm for clustering load curves and their influences on the clustering results are analyzed in this paper, the matching relations between normalization methods and clustering algorithms are obtained. Numerical examples show data normalization methods have different influences on clustering results using the same clustering algorithm.

Discretization is an important algorithm and considered to be a process of information generalization and data reduction. To avoid information loss and total number of cut point decrease after discretization of continuous attributes, based on multi-attribute discretization algorithm with good global clustering effects for selecting candidate cut points is proposed. The improved algorithm is combined with the advantages of clustering method and algorithm based on the importance of cut points. The experimental results show that the proposed algorithm can significantly decrease the number of discretization cut points and increase the predictive accuracy of the classifier than both.

The paper presents the novel approach to detection of some types of network attacks using web server logs. As the web server log files are just collections of strings describing users' requests to the server, the method of conversion informative part of the requests to numerical values has been proposed. The vector of values obtained as the result of web server log file processing is then used as the input to Self-Organizing Map (SOM) network. Finally, the SOM network has been trained to detect SQL injections and brute force password guessing attack. The method has been validated using the data obtained from a real data center.

A crisp image segmentation can be characterized in terms of the set of edges that separates the adjacent regions of the segmentation. Based on these edges, an alternative way to define a fuzzy image segmentation is introduced in this paper. In this sense, the notion of fuzzy image segmentation is characterized by means of a fuzzy set over the set of edges, which could in this way be understood as the fuzzy boundary of the image. Also, an algorithm to construct this fuzzy boundary is provided based on the relations that exist between the fuzzy boundary set problem and the (crisp) hierarchical image segmentation problem. Finally, some computational experiences have been included in order to show the fuzzy boundaries of some digital images.

Histogram modification techniques are an important tool for image enhancement. However, the efficiency of most of these techniques is limited due to the lack of mechanisms to differentiate between pixels that have the same intensity but belong to regions which are semantically distinct. In this paper, a method for histogram equalization, able to selectively modify each pixel in a digital image based on the context established by its neighboring pixels is proposed. This method emulates the decision–making process used by living organisms to recognize patterns, as described by the Gestalt theory.

HVAC (Heating, Ventilation and Air Conditioning) systems are important subsystems in buildings and it is necessary to control them in order to improve energetic efficiency. Conventional control syst ms have been used over the past years for controlling these systems with successful results. But classical control requires analytic models and HVAC based on fuzzy logic is proposed. The fuzzy control does not require the mathematical model and is able to incorporate the knowledge of an expert. Simulation results prove the feasibility of this solution.

This paper shows a comparison of different methods based on genetic algorithms in order to find a computational cost efficient path planning strategy for unmanned aerial systems (UAS). For that purpose, two different population generations and three crossover operators are proposed, comparing the computational time they require and the paths found. Results prove that it is possible to design a reliable and fast evolutive algorithm, capable of finding a sub-optimal solution without too high computational cost for a complex problem such as minimizing the path between two points for rotorcrafts.

This article presents, in contrast to more complex usual options, a simple and low computational cost solution for image stabilization. The solution is based on the minimization of frame differences and the use of a fuzzy logic supervisor. The fuzzy logic supervisor is designed as a control system for the stabilization to avoid undesired compensations. An analysis of the performance and viability of the system has been carried out and the results of this fuzzy approach are encouraging.

This paper presents an intelligent approach to the design and simulation of a conventional civil aircraft fuel system using fuzzy logic. The development is focused on the center of gravity balance problem in a plane, i.e., fuel re-distribution between several tanks during a complete operation (ground and flight). Additionally, the controller must guarantee the fuel feeding to the engines and allows the tanks refueling. Due to the problem characteristics we have selected fuzzy technique in order to design a controller with the appropriate features to accomplish all requirements. To make it more efficient, the rule base is reduced to simplify the design problem without loss of generality. Simulation results are encouraging.

In this paper, an intelligent control of the rotary inverted pendulum by fuzzy logic is presented. Specifically, the design consists of a Takagi-Sugeno fuzzy model to approximate the non-linear system to a succession of points where a linear system is described. A feedback gain is obtained that allows the stabilization of the inverted pendulum in a higher attractor than in the case of analytic Full State Feedback controller or Linear Quadratic Regulator.

This paper is a continuation of previous works on the semantic 2-tuples model, a representation model to deal with unbalanced linguistic terms sets. We propose to study how our semantic 2-tuples can help language processing since they offer a fuzzy semantic interpretation of words describing imprecise data. Thus, we propose two measures to catch the semantics of a set of words. We show the relevance of the measures in a use case where a lexicon has to be enriched.

The Iberian Electricity Market (MIBEL) is organized as a sequence of markets. Hourly marginal prices are obtained at the intersection of supply and demand curves. The price of electricity in the MIBEL is very changeable and the demand is a key variable for forecasting its final value for the 24 hours of the next day. Many different graphical techniques are proposed and implemented with R in order to explore, visualize and understand the behaviour of demand curves in the daily market along the time. This paper provides a graphical analysis by means of an easily reproducible and exportable automatization with R, a free software environment for statistical computing and graphics. Mibel 2011 and 2012 data are used for ilustrations. The results show the importance of the calendar effect, seasonality and trend as principal factors to take into account for posterior fases: modeling and forecasting.

Visual accessibility appears not to be the essential priority of artists when they produce an artwork. However, they do not choose colors randomly but depending on the message they want to deliver, which can be partially or entirely misunderstood by people with vision problems. This paper proposes to study the possibility of using expressive post-processing effects which can be applied on any kind of 2D image sequence and which allow to improve image perception for many kinds of visual impairments.

In this paper, we lay the groundwork for future research about a Polish intelligent tutor that would be able to teach learners for only a 14-hour-lesson, giving them the A1 level of the Common European Framework of Reference. The design and the implementation of the future software will be guided by linguists and psycholinguists.

This paper aims at giving thoughts about a proposal of a smart annotation regarding graphical documents. This research is dedicated to people with special needs, especially to blind or visually-impaired people. The idea is to propose a software that can annotate graphs with semantic descriptors (i.e. sentences) automatically.

The correct selection of performance metrics is one of the most key issues in evaluating classifier's performance. Although many performance metrics have been proposed and used in machine learning community, there is not any common conclusions among practitioners regarding which metric to choose for evaluating a classifier's performance. In this paper, we attempt to investigate the potential relationship among some common used performance metrics. Based on definitions, We first classify seven most widely performance metrics into three groups, namely threshold metrics, rank metrics, and probability metrics. Then, we focus on using Pearson linear correlation and Spearman rank correlation to investigate the relationship among these metrics. Experimental results show the reasonableness of classifying seven common used metrics into three groups. This can be useful for helping practitioners enhance understanding about the different relationships and groupings among the performance metrics.

The Development of Neuro-Evolutives (NEAs) Algorithms, designed specifically for evolving Artificial Neural Networks (ANNs) architectures, has aroused the interest of many researchers in the last twenty-five years. This paper shows that a direction to be followed in the development of NEAs more biologically plausible is to evolve ANNs through artificial development models. Whereas the analogous biological process should be viewed as a process of organization carried out by genetic information encoded in DNA and when followed will generate the final shape of the organs, including the brain. This research develops a biologically inspired methodology for automatic design of ANNs. With this goal using an artificial development system based on a parametric Lindenmayer with memory (allowing to incorporate aspects of organization, geometry and repetition of patterns in obtaining the topologies of neural networks), integrated to a Genetic Algorithm (GA) which simulates artificial evolution, allowing generate architectures of ANNs direct and recurrent with optimal number of neurons and appropriate topology. The technique was tested on XOR problem and in Classification Problem, some advantages of the proposed methodology is that it increases the level of implicit parallelism of Genetic Algorithm (GA) and seems to be capable to generate satisfactory neural architectures, reducing the project cost and increasing the performance of the obtained ANN.

Nowadays, many real world datasets are inherently overlapping clusters, and it is also a critical problem for the clustering analysis techniques to partition partially overlapping clusters. In this paper we propose a framework for overlapping clustering, and it consists of clustering, selecting and aggregation. In the clustering part, all kinds of clustering algorithms are suitable. In the selecting part, we propose a method called ConSim (confusion-similarity) to select some overlapping clusters. In the aggregation part, the selected clusters will be blended, and the overlapping objects will be found out. The framework is more flexible and powerful and it is demonstrated effective through experiments on datasets.

Co-clustering is an unsupervised machine learning technique, and it simultaneously clusters rows and columns for the input data matrix. In text co-clustering, the input data matrix is a high dimension and sparse matrix, and the traditional co-clustering ignores the similarity between word and word, and the similarity between text and word. In this work, we propose a text co-clustering using matrix block and correlation coefficient. In the first step, the matrix block and correlation coefficient are used to reduce dimension, and the relevant feature terms are merged as a hybrid feature term. In the second step, the text terms are clustered using K-means algorithm. And in the last step, the hybrid feature terms and text terms are iteratively clustered by the adjusting algorithms. Experimental results show that our algorithm is effective for a high dimension and sparse text matrix.

There are many university websites on internet. These websites can be considered as external information resources on the internet that academic institutions or students can use this external resources in order to improve their decision making process. It is therefore, very important and critical that the information of these external resources can be acquired precisely and on time. Most university web sites provide data in a semi-structured form on the internet. The combination of semi-structured data from different sources on the internet often fails because of syntactic and semantic differences. The access, retrieval and utilization of information from the different websites impose a need for the data to be integrated. Integration of web data is a complex process because of the heterogeneity nature of web data and thus needs some kind of a web data integration system. There are many types of heterogeneity and differences among university websites that makes data integration a difficult process (e.g., different data model, different syntax and semantics in schema and data instance level among web sources). In this paper, we recommend a system architecture for web data integration focusing on resolving the problems of semantic heterogeneity between university websites. We propose an ontology-based approach as a solution for the reconciliation of semantic conflicts between websites of universities and then develop a prototype of the proposed system for universities.

Most of the dynamic web applications generates HTML web documents that are deemed invalid as they do not adhere to the HTML standards defined under the World Wide Consortium (W3C). Though state-of-the-art web browsers are capable of rendering malformed HTML documents by correcting these errors discretely, most of the time such incidents poses compatibility issues which causes performance degradation for some applications. Various validation tools have been developed and are widely available across the Internet to address the mentioned issue, however, these tools only worked well for static web applications since it doesn't address the aggressive nature of dynamic web applications. Furthermore, validation tools targeting dynamic web applications available currently employ a static technique which is impossible in examining every single execution route possible. To resolve such issue, we have presented a novel framework for validating dynamic web applications through a set of heuristic rules which takes into consideration the syntax of several well-known server side languages such as PHP, Java EE and ASP.NET. As an extension to that paper, we now introduces a path finding and cross platform tools to validate HTML tags based on the server side language flow of execution.

Multi-follower tri-level (MFTL) decision making addresses compromises among three interacting decision units within a hierarchical system of which multiple followers are involved in two lower-level units. The leader's decision is affected not only by reactions of the followers but also by various relationships among them. The uncooperative relationship is the most basic situation in MFTL decision cases where multiple followers at the same level make individual decisions without any information exchange or share among them. To support such a MFTL decision, this paper firstly proposes a general model for the decision problem and then develops an extreme-point search algorithm based on bi-level Kth-Best approach to solve the model. Finally, a numerical experiment illustrates the decision model and procedures of the extreme-point search algorithm.

Social networking services (SNS) have had a rapid development. To explore the interpersonal node spatial distribution characteristics in SNS community, a fuzzy SNS community centrality analysis method is proposed, which combines graph-theory with related fuzzy approaches. A case study on two types of topic groups in Ren Ren Net is conducted. The study finds that centralization characteristics exist in the interpersonal node spatial distribution in SNS community and present significant difference among groups of different relation types. The findings can directly support the SNS management and development.

This paper studies manufacturer's logistics outsourcing decision in double channel supply chain. We respectively establish without free-rider manufacturers logistics outsourcing, logistics self-management and situation with free-rider of logistics outsourcing and self-management. And further we analyze the logistics outsourcing and self-management critical point. This paper mainly discusses market share and the cost of logistics system's influence on the manufacturer's logistics outsourcing decision-making. The results show that: manufacturer logistics outsourcing decision largely depends on the cost of logistics system; manufacturers and retailers exists both competition and cooperation relationship in the double channel of the supply chain, considering free-rider factors will change the point of manufacturers logistics outsourcing decision.

Detection systems of computer accesses are essential for information security. In this article we propose a classification system that combines two intelligent algorithms: Supervised Classification Systems, UCS, and Decision Trees, C4.5. The experiments were carried out using a dataset provided by Amazon, the Kaggle challenge. The system has been trained by dividing the dataset into subgroup. This training strategy has resulted more efficient than if the whole database is used as an only set. Results prove that the use of the proposed detection system provides higher classification accuracy and reduces the percentage of false positives in comparison to other classification techniques.

Ensemble of classifiers is a very popular method for online and incremental learning in non-stationary environment, as it improves the accuracy of single classifiers and is able to recover from drifting concept without explicit drift detection. However, current ensemble weighing methods do not consider the relationship between a test instance and each ensemble member's training domain. As a result, a locally correct ensemble member may be reduced weight unfairly because that its prediction result of an out of domain test instance is wrong. These inaccuracies will increases when there is a significant concept change. In this paper, therefore, we proposed a fuzzy online ensemble weighting method which takes the consideration of the degree of membership of each instance in each ensemble member and a modified majority voting method to improve the ability of ensemble on handling online classification tasks with concept drift.

Concept drift is a very pervasive phenomenon in real world applications. By virtue of variety change types of concept drift, it makes more difficult for learning algorithm to track the concept drift very closely. Learn++.NSE is an incremental ensemble learner without any assumption on change type of concept drift. Even though it has good performance on handling concept drift, but it costs high computation and needs more time to recover from accuracy drop. This paper proposed a modified Learn++.NSE algorithm. During learning instances in data stream, our algorithm first identifies where and when drift happened, then uses instances accumulated by drift detection method to create a new base classifier, and finally organized all existing classifiers based on Learn++.NSE weighting mechanism to update ensemble learner. This modified algorithm can reduce high computation cost without any performance drop and improve the accuracy recover speed when drift happened.

New trends in recommender systems face new challenges as group recommendation, in which users give their preferences over items and the system provides recommendations for a group of known users. In certain types of groups, it often occurs that several members do not agree on their preferences over some items so their inclusion in the group recommender system (GRS) may mislead the recommendation results. In this contribution a technique to detect and filter conflictive ratings before their use in the recommendation process is proposed and then its performance evaluated by using a well known recommendation dataset. The results show that rating filtering leads to improvements on GRSs performance.

Instance selection is an important pre-processing step in pattern recognition and machine learning. In this paper, we propose a novel instance selection method based on genetic algorithm for nearest neighbor (AGAIS_NN), which compose of three main parts: elitist strategy, adaptive probabilities of crossover and mutation, and fitness function. To validate the proposed algorithm, we compare AGAIS_NN with other classical instance selection methods. The experimental results show that our proposal is more effective and useful than other approaches.

Hesitant fuzzy sets are used to handle the situations where a set of values are possible in defining membership functions. In urban transformation problems usually there are multiple actors with different perspectives and they represent hesitant evaluations on subjective criteria. Hesitant fuzzy linguistic term sets (HFLTS) enable aggregating the different linguistic evaluations of different actors without loss of information. This paper proposes a hierarchical multiattribute method based on hesitant fuzzy linguistic term sets for prioritizing the urban transformation projects in Istanbul.

Computation of approximation in Dominance-based Rough Sets Approach (DRSA) is a necessary step for multi-criteria decision analysis and other related works. This paper presents a parallel approach for computing approximations of DRSA. Its feasibility is validated by a numerical example in this paper.

Seeing the business flows of service composition as the research object, this paper aims at bringing the security service provided by security components that deployed by system into composition service. Its major focus is in the premise of ensuring the accuracy of system business, to formulate the security efficiency that transitional operation composition of security service identifies system business, and to put forward a safeguard approach that reasonably restructure the security components targeting the security service which dissatisfies system requirements. By producing an appropriate redundant path of component service based on the existing execution records and logical structure, this approach will help reach the safeguard goals of business operation security in information system.

The control methods based on neighborhood systems can break down a dynamic and complex control process into a series of static and simple ones. They are available approaches for the control of automatic driving. According to a car's orientation relative to road, a neighborhood system for intelligent cars is presented in this paper. Methods to select some feasible neighborhoods for an automatic driving car are discussed, which are built by considering the changes of road edges and the relative speed to the front of cars or obstacles. All the methods are designed depending on data from available sensors of angle and distance. Each of the feasible neighborhoods could reflect some changes of the environment factors, so that it can be used to fuzzy predictive control. The effectiveness of the presented methods is simulated with a full size car on road.

The Virtual Organization (VO) concept has emerged as one of the most promising forms of collaboration among companies by providing a way of sharing their costs, benefits and risks, in order to attend particular demands. Although these advantages, VOs face several risks that need to be identified, measured, and mitigated through a well defined process. In this way, this paper proposes a hybrid DEA-Fuzzy method for analyzing risk in VO formation. This method assesses the level of risk present in a set of previously selected Service Providers (SPs) using Key Performance Indicators (KPIs), providing a way to helping decide on the VO formation.

In virtual reality-based 3D garment design, one important issue is to minimize the perceptual gap between real and virtual products in their static and dynamic representations so that they can be considered as the same by both designers and consumers. In this paper, we present a new method of online experimental design for quickly controlling human perception on virtual garments towards real products within a very few number of sensory tests. For each real product, this method uses the uniform design to generate the initial virtual fabrics then the principle of online active learning to sequentially create new virtual samples according to the evaluated similarity degrees of previous samples related to the real product. The proposed design of experiments will permit to identify the optimal values of the design parameters corresponding to the desired fabric. The criterion of data sensitivity is used to determine the most relevant design parameter on which we will enhance searches in the following step.

Existing research shows significant associations between online search volume and stock price on market level. Due to the individuality of each stock and limited attention of investors, how to effectively discover the association between search volume and the price on individual stock level is worth studying. This paper investigates the association pattern between search volume variation and stock price variation on individual stock level, and designs a so-called ATARII method, which can effectively discover qualified association rules between search volume and stock price in an after-temporal manner. Furthermore, real world data experiments are conducted on China's A-share stock market. Based on the discovered after-temporal associations, a new trading strategy is designed, i.e., ATARII-Trading, which obtains best cumulative return, significantly outperforming market level trading strategy, buy-and-hold strategy and random strategy as well as A-share 300 index.

The main goal of this paper is to expose the possibilities for applying of the new fuzzy methods for the evaluational modeling of credit risk, which is in its nature a composite quantity, and as such could be conveniently modeled through Interpolative Boolean Algebra. This approach to modeling allows expressing intensity measures of single components/properties of credit risk, as much as complex logical constructions made of basic components and their logical interactions, which altogether increases the possibilities of consistent mathematical articulation of the problem of aggregation of multicriterial aspects of credit risk into a single representative parameter.

This paper discusses the partial r-interdiction median problem for multi-sourcing supply systems (PRIM-MS). Compared with r-interdiction median problem (RIM), PRIM-MS mainly has three characteristics: (1) Limited capacity for each facility; (2) Partial interdiction; (3) Multi-sourcing property for supply systems. PRIM-MS falls into bilevel programming in the form of leader-follower decisions, where the attacker tries to degrade system performance the most by determining offensive resources invested while the user rebuilds supply-demand relationship to reduce the influences by interdiction. In this paper, we model the problem as a single level heuristic model based on two heuristic strategies. Experiments are carried out to testify the effectiveness of the heuristic model.

Selection of relevant parameters from a high dimensional process operation setting space is a problem frequently encountered in industrial process modelling. This paper presents a method for selecting the most relevant fabric mechanical parameters for each sensory quality feature. The proposed relevancy criterion has been developed using two approaches. The first utilizes a fuzzy sensitivity criterion by exploiting from experimental data the relationship between mechanical parameters and all the sensory quality features for each evaluator. Next an OWA aggregation procedure is applied to aggregate the ranking lists provided by different evaluators. In the second approach, a panel of experts provides their ranking lists of mechanical features according to their professional knowledge. Also by applying OWA, the data sensitivity-based ranking list and the knowledge-based ranking list are combined to determine the final ranking list and the final relevant mechanical parameters for a given sensory quality feature.

In this paper, we propose a self-adaption MapReduce framework based on granular computing (GrC), named gMapReduce. In the MapReduce model, the input data is partitioned into many data blocks which is the key step for the following parallel processing. The number of data blocks depends on the size of the block. It means the block size will affect the total running time. According to the proposed gMapReduce model, we design two algorithms, naive and advanced, for finding the appropriate granule. Both two algorithms can find the appropriate size of data block, thereby accelerating the running process effectively.

This paper investigates the use of RBF neural networks for software effort estimation. The focus of this study is on the design of these networks, especially their middle layer composed of receptive fields, using the Self Organizing Maps (SOM). An evaluation of the accuracy of effort estimation models that use either an RBFN construction-based on SOM, C-means or APC-III, is hence presented. This study uses the Tukutuku dataset.

This paper introduces an evolving neo-fuzzy neural network with adaptive feature selection approach in which candidate models with larger and smaller number of input variables than the current model are developed concurrently. The best amongst the current and candidate models is chosen at each step. The approach uses an incremental learning algorithm to simultaneously update the weights, to select the input variables, and evolve the network structure. Computational experiments concerning identification of a nonlinear process is performed to evaluate the method and to compare its performance against alternative evolving models. The results show that the extended evolving neo-fuzzy neural network with adaptive feature selection approach achieves higher or as high performance as alternatives evolving modeling methods.

Traditional patent classification schemes, which are mainly based on either IPC or UPC, are too complicated and general to meet the needs of specific industries. The paper proposes a dynamic classification method, the “user demand-driven patent topic classification”, aiming to a specific industry or technology area. In the paper, classification topics of the method are grouped into technical topic, application topic and application-technical mixed topic. Automatic process of the method using machine learning techniques is presented as well. A case study on the technology area of system on a chip (SoC) is conducted using machine learning techniques, validating the feasibility of the method. The experiment results demonstrate that automatic patent topic classification based on the combination of patents' metadata and citation information can obtain perfect performance with a greatly simplified document preprocessing.

In this paper we present a study aimed to define general openings and closing as specific morphological filters. Specifically, we study how much openings and closing we can be define just by using idempotent dilations and erosions.

Environmental quality plays a very important role in the human survival and development, and therefore environmental quality assessment must be effectively carried out. The paper presents an adjustment regional environmental quality assessment based on differences of group bearing capacity, the strategy will help people to distinguish environmental quality real impacts on people and help people pay attention to and protect environment quality.

Domain adaptation addresses the problem of how to utilize a model trained in the source domain to make predictions for target domain when the distribution between two domains differs substantially and labeled data in target domain is costly to collect for retraining. Existed studies are incapable to handle the issue of information granularity, in this paper, we propose a new fuzzy domain adaptation method based on self-constructing fuzzy neural network. This approach models the transferred knowledge supporting the development of the current models granularly in the form of fuzzy sets and adapts the knowledge using fuzzy similarity measure to reduce prediction error in the target domain.

This paper describes a methodology based in Fuzzy Sets Theory to elaborate a model that studies the interaction between the prey, Aphis glycines (soybean aphids), and the predator, Orius insidiosus. The aim of this investigation has been to develop a simple and specific methodology to make a decision on the control of this prey.

Logistics plays a vital role in the economies. One of the attempt to measure logistics performance at national level is the Logistics Performance Index (LPI) published by the World Bank Group. In fact this study argues that there is a close relationship between global competitiveness level of a country and its logistics performance level. On this way, it aims to analyze the logistics competitiveness of a country from a national competitiveness perspective using an analytical neural network and cumulative belief degrees approach. LPI and World Economic Forum's competitiveness pillars are used for this purpose. The methodology is used to analyze the Turkey's logistics performance in order to develop suggestions for improving its current level.

This paper describes a new way of constructing a visuomotor map that is applied in a manipulator robot to accomplish a ballistic reaching task. Ballistic reaching is employed to move the robot hand to a visible and optimal position, once the robot arrives to this position a visually guided task (visual servoing or a manipulation task) could be executed. The visuomotor map has been implemented using SOM neural networks in a simulated puma 560 robot. The visuomotor map deals with the visibility of the robot hand (end-effector) problem. So the robot can start with its end-effector in any random position of the workspace. A Cartesian controller triggered by a visuomotor map will find a suitable and visible position for the end-effector. Results and simulations are shown to demonstrate the applicability of our proposed model for a ballistic reaching task.

Hybrid morphological/linear neural networks combine morphological with linear operators. In this paper, we introduce a feedforward artificial neural network representing a hybrid fuzzy morphological/linear perceptron called fuzzy dilation/erosion/linear perceptron (F-DELP). Following Pessoa's and Maragos' ideas, we apply an appropriate smoothing to overcome the non-differentiablity of the fuzzy dilation and erosion operators employed in the proposed F-DELP models. Then, training is achieved using a traditional backpropagation algorithm. Finally, we apply the F-DELP model to some well-known classification problems and compare the results with the ones produced by other classifiers.

Traditional tobacco intelligent sensory evaluation has no high accuracy due to the limitation of chemical components detection technologies. A new approach discussed in this paper utilizes the near-infrared spectroscopy (NIRS) to build the model of sensory quality instead of the analysis of chemical compositions. The results indicate that the NIRS models have greater ability to treat complex tobacco sensory analysis. The NIRS sensory evaluation which is fast, efficient and environmentally-friendly will become more prosperous in this field.

Learning the structure of a Bayesian network classifier (BNC) encodes conditional independence assumption between attributes. One major approach to mitigate BNCs's primary weakness (the attributes independence assumption) is the locally weighted approach. And this type of approach has been proved to achieve good performance for NB, a BNC with simple structure. However, we do not know whether or how effective it works for improving the performance of the complex BNCs. In this paper, we carry out an systematically experimental analysis to investigate the effectiveness of locally weighted method for complex BNCs measured by the area under the ROC curve ranking (AUC). Experiments on 36 benchmark data sets in Weka system demonstrate that although locally weighting significantly improve the performance of NB (a BNC with simple structure), it could not work well on BNCs with complex structures.

The main purpose of this research work is to introduce and study the concept of F-transform based on generalized residuated lattices.

The following sections are included:

• Summary
• References

The following sections are included:

• SUMMARY
• References

The following sections are included:

• AUTHOR INDEX
• ACKNOWLEDGEMENTS