In this paper, fuzzy c-lines for data with tolerance is proposed. First, Fuzzy c-Lines without tolerance and Fuzzy $c$-Means for data with tolerance are introduced. Second, an optimization problem is proposed for Fuzzy $c$-Lines for data with tolerance. Third, Its KKT conditions are solved into an alternative optimizing algorithm is proposed. Forth, some numerical examples are shown. Last, this paper is concluded.

View full abstract

In this paper, we propose a kernel aggregation operator and present an application of this operator to multidimensional scaling. The purpose is to improve the performance of the multidimensional scaling by using our proposed aggregation operator. The idea is based on a combination of statistical metric space and kernel method.

View full abstract

We propose a clustering model based on probabilistic dissimilarity between clusters by assuming a probabilistic distribution for each cluster. The feature of this model is that we calculate the dissimilarity directly by using a result of clustering and then estimate the final clustering result. The merit is that we can obtain the asymmetric feature of clusters explicitly.

View full abstract

This paper proposes the modifications of the fuzzy c-means (FCM) based classifier. The FCM classifier uses covariance structures to represent flexible shapes of clusters. Despite its effectiveness, the intense computation of covariance matrices is an impediment for classifying a set of high-dimensional data. In order to tackle with this problem, we proposed a way of directly handling high-dimensional data in the FCM clustering. The third type of the FCM classifier proposed in this paper is the relational classifier. The classifier treats relational data instead of object data. All these three classifiers are equivalent when the dimensionality of feature vectors is not very large, and the relational data are obtained based on Euclidean distances between object data. Numerical test results reveal the good classification performance of the triplet.

View full abstract

In this paper, we propose two variants of Fuzzy c-Means (FCM) clustering algorithm. Both models are extended to admit the case where the dissimilarity measure on pairs of numerical vectors includes p-norm to the qth power. One is the algorithm using Newton method, and the other is the algorithm for the case of p = 2, using alternative optimization.

View full abstract

Clustering is one of the unsupervised classification and fuzzy c-means (FCM) is one of the typical technique of fuzzy clustering. Endo et al. have introduced the concept of tolerance and constracted the algorithm of FCM for data with tolerance (FCM-T) to handle uncertainties with data. In the algorithm, the constraints for tolerance vectors are used. In this paper, we will try to get rid of the constraints by introducing the penalty term instead of there. On the other hand, the dissimilarity of FCM is defined as the squared L2-norm. Moreover, L1-norm based methods are also constructed. L1-norm methods can calculate results rapidly. In this paper, we will consider L1-norm based FCM.

View full abstract

Recently, studies for semi-supervised classification has been developed. This method is focused because of obtaining the better classification results and decreasing calculation cost. In this paper, a new semi-supervised clustering is proposed by applying some laveled membership grade to fuzzy c-means.

View full abstract

We have proposed L2 or L1-norm based tolerant fuzzy c-means clustering (TFCM) from the viewpoint of handling data more flexibly. This paper presents a new type of tolerant fuzzy c-means clustering with L1-regularization. The L1-regularization is well-known as the most successful technique to induce sparseness. The proposed algorithm induce the sparseness for tolerance vector. First, tolerant fuzzy c-means clustering is introduced. Second, the optimization problems with L1-regularization are solved. Third, a new clustering algorithm is constructed based on the explicit optimal solutions. Finally, the effectiveness of the proposed algorithm is verified through numerical examples.

View full abstract

It is very difficult to construct algorithms which gives regression model for both of independent and dependent data with uncertainty which is represented by interval. The reason is caused by difficultiy of the mathematical analysis. In this paper, we will solve the problem using a concept of tolerance and iterative optimization and construct a new hard c-regression model for data with tolerance.

View full abstract

First, we formulate independent and dependent data with uncertainty that is represented by interval using a concept of tolerance as data with tolerance. Next, we propose a new fuzzy c-regression model for data with tolerance.

View full abstract

A standard and explicit mapping into a high-dimensional feature space is proposed for kernel-based data analysis. It will be shown tht using this method, no disadvantage will occur while many kernel methods will be simplified. An application to text analysis will be shown.

View full abstract

Presented in this article presents are two studies on the web semantic interfaces of commercial sites, one pertaining to textual structures and the other to eye-tracking records. For the former, the web pages of the top two skincare brands were compared, while the top pages of commercial cites were analyzed in the latter. The common strategy was extraction of core neighborhoods on the basis of centrality indices developed in network analysis with the aid of layout algorithms. The brands exhibited quite contrasting network structures. Yet, both shared the same core term "plump," probably reflecting the common concern about the skincare. What emerged from the eye-tracking records were fairly good match between the core-peripheral nodes and their corresponding locations on the screen. Potential use of the present approach was discussed with caveat.

View full abstract

Under construction

View full abstract

This paper proposes an FCA technique for representing classifications, which is a use of the apposition with the nominal scale on the objects. It removes the hierarchical order between the object concepts, while it preserves the hierarchical orders between the attribute concepts and between the object concepts and the attribute concepts. As an example, it is applied to cladistic analysis.

View full abstract

A sinple function can be represented in three different forms using the definition function. Based on these representations, we have three different types of fuzzy integrals. Here we consider how the integrals having differnt meanings come out from the same single function.

View full abstract

Choquet integral models are useful global evaluation models including interaction effects among evaluation items. Evaluation rules may be changed whatever input values stisfiy the reference points. Choquet integral models with constants enable to deal with the situations. The proposed models regard the constants as the reference points.

View full abstract

A new integral for non-additive measure is proposed. We construct a model with this integral for evaluation problem and verify its feasibility applying a real data comparing with the multiple linear regression analysis and the Choquet integral model.

View full abstract

This paper provides a new value (solution concept or allocation rule) of cooperative games via posets induced by graphs. Several values in a graph-restricted communication situation have been proposed or introduced by Myerson, Borm, and Hamiache... However, these values have been subjected to some criticisms in certain types of games. The value proposed in this paper withstands these criticisms. Moreover, these existing values have been defined only in situations represented by undirected graphs, while the notion of the value proposed in this paper can be extended to situations represented by directed graphs.

View full abstract

In this paper, we propose an interactive genetic algorithm applying paired preference test that obtains evaluations of individuals from multiple people. In this paper, the effectiveness of the proposed technique was verified by a simulation using multiple evaluation agents in place of real evaluators. In the simulation, when 100 the evaluation agents participated in a vote, we can reduce necessary votes for the evaluation of the solution 25% than conventional technique. Moreover we confirmed that the proposed technique was more effective than the conventional technique from the viewpoint that efficiency of the tournament competition.

View full abstract

The iterated prisoner's dilemma (IPD) game has been frequently used to examine the evolution of cooperative behavior among agents in the area of evolutionary computation. A number of factors are known to be related to the evolution of cooperative behavior. One well-known factor is spatial relations among agents in a grid-world. Such a spatial IPD game has a neighborhood structure which is used for local opponent selection in the IPD game and local parent selection in genetic operations. Another important factor is the choice of a representation scheme to encode each strategy. Whereas it is known to be important, a mixture of different representation schemes has not been examined for the spatial IPD game in the literature. In this paper, we introduce the use of a mixture of different representation schemes and the use of two neighborhood structures in the IPD game. Through computational experiments, we examine the effects of these specifications to the evolution of cooperative behavior.

View full abstract

Recently evolutionary multiobjective optimization (EMO) algorithms have been actively used for the design of accurate and interpretable fuzzy rule-based systems. This research area is often referred to as multiobjective genetic fuzzy systems where EMO algorithms are used to search for a number of non-dominated fuzzy rule-based systems with respect to their accuracy and interpretability. The main advantage of the use of EMO algorithms for fuzzy system design over single-objective optimizers is that multiple alternative fuzzy rule-based systems with different accuracy-interpretability tradeoffs are obtained by their single run. In this paper, we discuss the ability of EMO algorithms as multiobjective optimizers to search for Pareto optimal or near Pareto optimal fuzzy rule-based systems. More specifically, we examine whether EMO algorithms can find non-dominated fuzzy rule-based systems that approximate the entire Pareto fronts of multiobjective fuzzy system design problems.

View full abstract

In optimization process of genetic algorithm application, early convergence often happens through iterations of the generation alternation. In order to avoid such situation, individuals in GA should be sufficiently varied in whole generations(Diversity maintaining). Distributed genetic algorithm(DGA) is a promising approach for the diversity maintaining as well as acceleration of the calculation. This paper presents a study on topogy of each group(island) in construction DGA based optimization system. We discuss the performance especially resulting from the difference of topologies structured as Small-wold network in DGA, through numerical experiments of scheduling problem, in terms of accuracy and diversity of the solutions.

View full abstract

Shogi game is similar to Chess and very popular in Japan. Computer Shogi programs are still in developing to defeat the professional human players. One of the main problems exists in estimating the circumstances of the game phases. It is said that there are three phases in the Shogi game, so called, opening, middle and endgame phase. The appropriate strategy to be selected differs depending on the proceeding phase. The professional human players classify states of the game phases properly. In this paper, we have carried out some experiments to extract human palyers' strategies on shogi game from game records using growing SOM. The results show the promising feature of the proposed method.

View full abstract

Game competitions in FUZZ-IEEE and CEC (Congress on Evolutionary Computations) have been held to develop the techniques of computational intelligence in the real world. In this paper, we tried to acquire the human skill using human players' logs. The Self-generating and Self-organizing maps are evoluved through learning the logs.

View full abstract

In this paper, we propose a neural network which learns knowledge from natural language documents and analogize. In proposed network, firing patterns of neurons are memorized in Form Inference Memory and Meaning Inference Memory. When a similar firing pattern is appeared, a memorized pattern is retrieved from Memories. Analogical inference is allowed by this processing. In learning process, natural language documents are analyzed by Japanese dependency structure analyzer, CaboCha. The results are used in network connection learning. By these processing, the proposed network can learn and analogize. We confirmed the effectiveness of the proposed network.

View full abstract

We developed a System for Noun Concepts Acquisition from utterances about Images, SINCA in short. It is a language acquisition system without knowledge of grammar and vocabulary, which learns noun concepts from user utterances. At present, we aim to develop a system which acquires adjectives using a method similar to SINCA's one. In this paper, we describe the system which acquires color adjectives and an evaluation experiment.

View full abstract

The safety of usage of medicines have been a focus of attention to ensure medical safety by right use of right medicines. Though usage regimen, which shows the information about proper usage, is described in package inserts of the medicine as an official document, it is insufficent to utilize this information for the prevention of medical accidents. In fact, the data in package inserts are described in sentences, not in table as structured data. In order to extract information of dosage regimen from package insert data, we have to leverage natural language techniques. Authors have proposed a `word-link' method to find common sentence structures embedded in large number of (short) documents. In this study, we apply the extended version of our method to find the structures of information contained in dosage regimen description. The result will help to define the data schema of dosage database.

View full abstract

Analysis of the impresshion of the name of the color - comparison of impression for color terms in Japanese and English-

View full abstract

Ishikawa Prefecture is located in the Hokuriku region in Japan. One of the problems of the tourism in Ishikawa is to increase the number of tourists from foreign countries. In order to solve this problem, it should be necessary to provide foreign tourists with "language service." In this study, in order to understand a state of language service to foreign tourists, we investigated what linguistic characteristics can be found in English pamphlets at the Komatsu Airport and the Toyama Airport.

View full abstract

This article presents random fuzzy portfolio selection problems including the investor's subjectivity. The proposed models are formulated as stochastic and fuzzy programming problems, and so in order to solve these problems analytically, we introduce stochastic and fuzzy chance constraints and perform deterministic equivalent transformations. Furthermore, introducing some numerical examples, we compare our proposed models with standard portfolio models and consider the effects for the investor's subjectivity.

View full abstract

This paper proposes a new location model of competitive facilities, e.g. shops and stores, on a network. In the model, demands for the facilities are represented as flows on the network, e.g. customers on main streets. Since such demands are usually uncertain for the reasons of weather, economic conditions, etc, the amount demanded with the flows are represented as fuzzy numbers. Then, the competitive facility location problem is formulated as a fuzzy programming problem. Because the problem needs numerous computational times for the case that the scale of network is large and there are many competitive facilities, its solution method based upon tabu search algorithm is proposed. The efficiency of the solution method is shown by applying to several examples of competitive facility location problems.

View full abstract

In this research, focusing on nonlinear integer programming problems, we propose an approximate solution method based on particle swarm optimization proposed by Kennedy et al. And we developed a new particle swarm optimization method which is applicable to discrete optimization problems by incoporating a new method for generating initial search points, the rounding of values obtained by the move scheme and the revision of move methods. Furthermore, we showed the efficiency of the proposed particle swarm optimization method by comparing it with an existing method through the application of them into the numerical examples. Moreover we expanded revised particle swarm optimization method for application to nonlinear integer programming problems and showed more effeciency than genetic algorithm. However, variance of the solutions obtained by the PSO method is large and accuracy is not so high. Thus, we consider improvement of accuracy introducing diversification and intensification.

View full abstract

Bala and Goyal (2000) have constructed a mathematical model of network formation leading a center sponsored star network to be strict Nash Equilibrium. Basing on the model, Berninghaus et al. (2007) have conducted the laboratory experiments using human subjects . The result of the experiments indicates that peripheral sponsored star networks which are predicted as the strict Nash Equilibrium are not always stable. The partial subjects deviate from the Equilibrium strategies and several kinds of star networks with different central player are formed. One of the reasons why such process of the network formation is observed, the mechanisms of human are not always rational but also make decision through trail and error, occasionally. In this paper, we analyze the cause of the deviation from the Nash equilibrium by using agent-based simulation system.

View full abstract

In this paper, we focus on fuzzy random multiobjective 0-1 programming problems. Introducing fuzzy goals for objective functions to consider the ambiguity in the judgment of the decision maker, we transform the minimization of objective functions in the problems into the maximization of the satisfaction degree of each fuzzy goal. Since those degrees are random variables, we reduce the transformed problems to deterministic multiobjective 0-1 programming problems using probability maximization. In order to derive a satisficing solution for the reduced multiobjective problems, we propose an interactive fuzzy satisficing method through interactions with the decision maker.

View full abstract

Flexible Representation of Quantum Images (FRQI) is proposed to provide a representation of images that enables efficient preparation and image processing operators on quantum computers. FRQI captures colors and their corresponding positions in an image into a quantum state. The preparation with polynomial simple operations for FRQI turning a quantum computer from the initialized state to the FRQI state is shown by the proposed Polynomial Preparation theorem. The Quantum Image Compression (QIC) algorithm reduces simple operations used in the preparation process based on the minimization of Boolean equations are from same color groups of positions. Quantum image processing operators on FRQI based unitary transforms dealing with only colors, colors at some positions and the combination by the quantum Fourier transform of both colors and positions are addressed. The experiments for the storage and retrieval of images using FRQI are implemented in Matlab. The compression ratio of QIC among same color groups ranges from 6.67% to 31.62% on the Lena image.

View full abstract

A mechanism for reducing reverse traffic in the Fuzzy Logic based dual explicit congestion notification mechanism is proposed for satellite networks. A linguistic hedge based BECN activation function is employed in order to ensure that reverse traffic is generated only when congestion is extremely high. A mechanism that resets the BECN capability notification bit in the IP header is also employed in order to stop the router from generating more reverse traffic within a roundtrip time (RTT) after the sender has already responded to the congestion signal. This algorithm is compared with the basic fuzzy logic based dual explicit congestion notification mechanism and the fuzzy logic based dual explicit congestion notification mechanism, which uses a router computed RTT based decay function for reverse traffic minimization. Performance evaluation is carried out by using both web-traffic loads and a varying number of FTP flows. Simulation results show that the proposed approach reduces BECN Internet Control Message Protocol (ICMP) Source Quenches significantly by 75.4% while the RTT based decay function approach achieves a 43% reduction; both approaches do not degrade the packet loss rate and the link utilization.

View full abstract

The development of a top-down, hierarchical modeling architecture called a "knowledge tree" is introduced. Such trees comprise of nodes with local models that approximate data at the specific nodes. By clustering, a tree is unfolded successively. The novelty of this approach lies in the ability to refine the clusters and models in a step-wise fashion compared to single global rule-based structures such as the Takagi-Sugeno model. More so, local models at each node augment the overall architecture with high functionality compared to a decision tree whose nodes compute some broad classification rules. Experimentally, using synthetic and real-world data sets, compared to a single global model, a hierarchical tree is built faster and more efficiently while retaining high accuracy. Being hierarchical and step-wise, knowledge trees can be promising in large-scale applications (data of high dimensionality).

View full abstract

Smooth Outline Tracing of Car Using Fuzzy Inference

View full abstract

This paper presents polynomial fuzzy observer design for polynomial fuzzy systems. We derive observer design conditions in terms of sum of squares (SOS). Moreover, SOS conditions for simultaneously designing the polynomial fuzzy observer and polynomial fuzzy controller based on estimated state are derived. A design example illustrates the validity of the proposed method.

View full abstract

This paper deals with stability analysis and control design for a class of delay-dependent nonlinear systems when using the T-S fuzzy model in consideration of uncertainties in the system. As a result, it arrives at a PDC controller in terms of the LMIs, which involves less information about the delay times. In order to make the LMIs less conservative, the controller also makes effort to use terms that usually is ignored but can help the LMIs be easily satisfied.

View full abstract

This paper presents a concept to design piecewise nonlinear control systems using feedback linearization approaches. It was difficult to calculate an optimal solution of the stabilizing controllers because the conditions are represented as bilinear matrix inequalities. Our proposed approach is easy to design the stabilizing controllers because the design method is represented as the simple linear algebra equations. We apply the our proposed methods to nonlinear system controls. The illustrated examples are given to show the validity of the proposed methods.

View full abstract

Based on the piecewise Lyapunov function, this study proposes a switching fuzzy controller, which switches depending on the Lyapunov function, to get relaxed stabilization conditions for the continuous T-S fuzzy system. The relaxed conditions are bilinear with the s-procedure parameters, therefore the particle swarm optimization (PSO) algorithm is utilized with the LMI tool to solve the relaxed conditions. Two simulation examples are given to show the relaxation and effectiveness of the proposed method.

View full abstract

This paper considers a control design for fuzzy systems with distributed delays. The system is usually modeled a continuous-time system while system measurements are taken each sampling instants. First we show that the closed-loop system with such a control input becomes a fuzzy system with discrete and distributed delays. Sufficient conditions for the closed-loop system to be asymptotically stable are given in terms of linear matrix inequalities(LMIs). Based on the conditions, we propose a control design of stabilizing state feedback controllers for a fuzzy system with distributed delays. Finally, numerical examples are given to illustrate our design methods.

View full abstract

This paper reports on the performance of the fuzzy c-means based classifier (FCMC) adopting the change rates of mixing proportion of clusters as the free parameters. UCI benchmark datasets are used to evaluate the performance. FCM classifier in combination with standard 10-CV procedure or resubstitution (i.e., 1-CV) procedure for parameter selection achieves good test set performance compared to k-nearest neighbor classifier. Randomized test sets performance of the classifier is comparable to that of the support vector machine (SVM) reported in the literature.

View full abstract

Pawlak's rough set theory handles imprecision and uncertainty without the need for preliminary or additional information regarding analyzed data. Using rough set theory, knowledge extracted from a database may be constructed and generated in the form of comprehensible decision rules. However, increasing the number and complexity of decision rules could make their analysis and validation by domain experts difficult.In the present paper, we consider this problem and propose rough-set-based multi-valued decision diagrams (R-MDD) as an approach to visualize decision rules.

View full abstract

Knowledge discovery in databases is the non-trivial extraction of potentially useful information in databases. Many of conventional methods extract knowledge from data in one place. However, it is often the case where data is generated and stored in two or more places. In this paper, we propose a method to extract fuzzy rules in an entire decentralized data by transferring only values necessary for the calculation process, without collecting all decentralized data. The characteristics of data in each place can be expressed by comparing the rules for the entire data with those for each data. We illustrate a result for experiments using Iris data by R.A.Fisher.

View full abstract

Car racing game is a competition game of computer programs in IEEE 2007 Car Racing Copmpetition, where two car agents compete with each other for taking waypoints in a two dimensional plane. If the current waypoint is taken, the next waypoint become the current waypoint. After each action, the car agent can get some information on itself, the other car agent, the current waypoint and the next waypoint. In this paper, we evaluate how good are states of the car agents with the fuzzy sets about its speed and the distance and angle to the current waypoint from the information. We propose a method that compares its own and the other states and decides which waypoint is better to go. We illustrate this method is higher in performance in comparing the number of the winning games than the programs in Competition that do not use the fuzzy sets.

View full abstract

In this paper we propose the exponential fuzzy numbers in Analytical Hierarchy Process (AHP), employing the fuzzy concept. The objective was determined a method of deriving the weights of criteria and their alternatives. Next, the influence of fuzzy is examined by using the membership function. Specifically we use the exponential fuzzy numbers for the pairwise comparison scale. Same experiments with AHP were conducted using the pairwise comparison scale and the new pairwise comparison scale of the method was examined.

View full abstract

Rough set theory was proposed by Z.Pawlak in 1982. This theory can mine knowledge through a decision rule from a database, a web base, a set and so on. The decision rule is used for data analysis as well. And we can reason for an unknown object using the decision rule. In this paper, we apply the rough set theory to analysis of time series data. We acquire knowledge for tick-wise price fluctuations.

View full abstract

In this paper, the fuzzy reasoning demand forecasting model is examined by using the management data and meteorological factor during year.And, consumer's demand forecasting model is constructed by using meteorogical factor in statistically clarifying the relation between these. Moreover, the fuzzy reasoning model to whom demand is forecasting by using the factor obtained according to the analytical result is constructed.

View full abstract