This paper aims at evaluating the method for reduction of Interactive Evolutionary Computation (IEC) user's fatigue using a rating-scale mapping. The rating-scale mapping method is used to calculate absolute values from relative evaluation values that IEC user rates in each generation and train IEC user's evaluationcharacteristics effectively for accelerating IEC convergence. We first evaluate the method using seven benchmark functions as pseudo IEC users and show the method accelerate IEC convergence. The results of the simulation show that the convergence speed of an IEC using the proposed absolute rating predictor is much faster than using a conventional one. Next, we evaluate the method through subjective test using IEC based CG lighting system in order to prove that the proposed predictor is effective in reducing the user fatigue. The results show that most of the subjects using the predictor can find an optimum solution faster than using the conventional one. Finally, we discuss the results for further improvement.

View full abstract

The make-up that slightly changes the color and the shape of parts of woman's face changes the impression of the person who sees. In this paper, the make-up was targeted to the lip, the eye shadow, the teak, and nose shadow. First, these make-up conditions were examined and whether the image was able to be measured automatically was examined by an experimental system. Next, the make-up assumed to be "It was pretty" and "beauty" was made faces, and it experimented what impression those make-ups had given the woman. Consequently, the possibility of the measurement of the image of the make-up condition was found. And, some investigation results of the relation between the make-up and the impression were obtained.

View full abstract

In multi-agent systems, communication among agents plays an important role for cooperative behavior. In this paper, we investigate the effect of direct communication in a multi-agent system, Ant System (AS). Ant agents in AS usually perform indirect communication through pheromones. We introduce the mechanism that ants communicate directly with neighboring ants so that each ant's information can be exchanged. As experimental results, the AS with direct communication acquired better solutions than conventional AS. If the ants have excessive range of direct communications, however, the search performance becomes worse. We showed that local communications are effective in order to search solutions while striking a balance between centralization and diversification of the ants' action. For the effective search, it is important to set the range of the direct communication appropriately.

View full abstract

The Particle Swarm Optimization method is one of the most powerful optimization methods available for solving global optimization problems. However, knowledge of autonomous and adaptive strategies for tuning the parameters of the method for application to large-scale nonlinear non-convex optimization problems is as yet limited. This paper describes an autonomous and adaptive strategy for tuning the parameters of the PSO method based on some numerical analysis of the behavior of PSO. The proposed autonomous and adaptive tuning strategy is based on self-tuning of the parameters of PSO, which strategy utilize the information about the frequency of an updated global best of a swarm. The feasibility and advantages of the proposed autonomous adaptive PSO algorithm are demonstrated through some numerical simulations using two typical global optimization test problems.

View full abstract

It is well known that Particle Swarm Optimization (PSO), which was originally proposed by J. Kennedy et al., is a powerful algorithm for solving unconstrained and constrained global optimization problems. Appropriate adjustment of its parameters, however, requires a lot of time and labor when PSO is applied to real optimization problems. In this paper, we point out that giving diversity to the parameter of each particle enables PSO to solve various problems.And we propose an algorithm that has diversity of particles and an adaptive strategy for tuning a parameter.Some numerical simulations were carried out in order to examine the search ability of the proposed approach.

View full abstract

A swarm behavior has been simulated by many researchers. On the other hand, a research of a swarm control by an external force, such as a control of a flock of sheep by shepherd, is not popular. In this paper, we try to control a swarm by a simple rule sets. We assume that the system controls a behavior of fifty sheep by two sheep dogs. The proposed method is confirmed by computer simulations.

View full abstract

A new relevance feedback based on a tree structure with visual keys is proposed to reflect the user's sense in the region based image retrieval (RBIR) system employing query with visual keys. For relevance feedback, the region image which is obtained from a feedback image is linked to the correspondent leaf node in a tree structure, but it is not effective to add the leaf node to the query like in the usual relevance feedback because of the small increase in the number of the retrieved images. In the proposed relevance feedback, the device in which an upper node from the leaf node is added to the query is implemented. Image retrieval experiments using 1,000 test images in COREL gallery show that an average retrieval precision is improved from one of conventional relevance feedback using leaf nodes by the proposed method using upper nodes, and the effect of the relevance feedback is confirmed from an obtained recall-precision curve.

View full abstract

The differences of understanding of words among individuals often lead to miscommunication or misimpression. These differences are dependent on individual subjectivity. This paper proposes a method that visualizes the relationships between two subjects' impression words employed to evaluate the same objects in Semantic Differential (SD) method. The proposed method constructs a visualization space by applying the Principal Component Analysis (PCA) and Orthogonal Procrustes Analysis (OPA) to the evaluation data. This visualization space enables the users to compare relationships between two subjects' impression words. This paper employs the design templates of Microsoft PowerPoint 2002 as the objects of the evaluation by the SD method. It shows the proposed method can visualize the relationships of each subject's impression words. It also shows that the proposed method, which enables the users to make comparisons of a lot of impression words between individuals, creates more suggestive results than the conventional method.

View full abstract

This paper aims at the construction of the system that generates story texts from some pictures and transforms a theme music based on impressions of generated stories. Inputs of the present system are information on some pictures and an original theme music. And outputs of this system are story texts and the transformed theme music reflecting impression of generated story scenes. This paper also describes the evaluation experiments to confirm whether the present system is able to transform the theme music reflecting impressions of generated story scenes appropriately or not.

View full abstract

A method of representating multibar figures and their subclass torus figures is proposed, and an automatic drawing system for torus figures based on graph and knot theory is presented. The multibar type graph that is an abstract concept of multibar figures is defined by the junction graph that represents the connections of the lines. The link structure is defined for torus type graph to be able to classify torus figures and to show geometrical constraints with edges and faces of them. The presented system and theorems give a basic tool for experiments on visual psychology and development of intelligent recognition systems.

View full abstract

In this paper, the logic of vagueness and Fuzzy logics are dicussed.

View full abstract

In these 40 years there has been much development on fuzzy set theory especially in the application area, but on basic theory there has not been a uniform and consistent theory which cover most part of fuzzy set theory. It is partly because that fuzzy sets were not defined exlicitly in the pioneering paper of Zadeh. But the primary intention of Zadeh is considered to distinguish the notion of fuzzy set from its membership function and to develop a theory of fuzzy sets as an extention of usual set theory. Now our natural interpretation is to give a satisfactory explanation by giving a model of fuzzy set theory on which ordinary set theory can be naturally extended. Here we present the basic part of our natural interpretation of fuzzy sets. We interpret fuzzy sets in a cumulative Heyting valued model for intuitionistic set thoery. With the interpretation various notions and properties in fuzzy set theory can be acquired consistently. As far as fuzzy set theory is considered as an extension of usual set theory, this interpretation seems to be most natural.

View full abstract

This paper proposes analogy between fuzzy concepts considering situations. A fuzzy concept in a situation is defined as a 2-tuple composed of a fuzzy set and a binary relation, which is called a situation-dependent fuzzy concept. An analogy and a weak analogy are defined as a bijection between fuzzy sets expressing situation-dependent fuzzy concepts in different situations.Some theorems say that an analogy and weak analogies exist between fuzzy sets expressing situation-dependent fuzzy concepts in any situations and there are order-isomorphism. This means that a fuzzy set expressing a situation-dependent fuzzy concept in some situation is obtained by a fuzzy set expressing the same concept in another situation and an analogy or a weak analogy.

View full abstract

[in Japanese]

View full abstract

This paper generalizes the concept of inclusion-exclusion coveringto the inclusion-exclusion family of subsets, which is not necessarily a covering, and describes its mathematical properties. Especially, the class of inclusion-exclusion families of subsets of a finite set $X$ is isomorphic to the free upper-bounded distributive lattice generated by $X$, where an upper-bounded lattice is a lattice with the largest element. In addition, the paper proposes inclusion-exclusion family models, provide variable selection and the adjustment of qualitative strength of additivity.

View full abstract

It is well known that fuzzy mapping plays an important and fundamental role in fuzzy analysis. In this paper, we put forward the concept of directional derivative of fuzzy mappings from R into E , and discuss some properties directional derivative.

View full abstract

Ordinal fuzzy switching functions are calculated by max, min and (1-x) operators.Logical Choquet integrals are calculated by Choquet integral instead of max-min calculations.Logical Choquet integrals satisfy lots of good properties such as the complementary laws, monotonicity when the function is monotonicity with binary inputs and so on.To demonstrate the properties, 3 examples, `Fuzzy XOR', `friendship and power' and `Evaluation of 3 subjects' are given.

View full abstract

Opposite Sugeno integral with normal measure as an image filter is investigated. Initial image is assumed to be an independently uniformly distributed gray image. The image after filtering was also uniformly distributed, but it had broad self-correlation.

View full abstract

The reasoning result is controlled by reasoning algorithm, accuracy of knowledge, and usage of knowledge. We will call reasoning algorithm, knowledge, and usage of knowledge three elements of the reasoning.In this report, a concept of knowledge radius is proposed, and how to use knowledge is discussed based on the knowledge radius to decrease the reasoning error margin.

View full abstract

A detection algorithm for dynamic color image taken by CCD camera in the real surveillance system is proposed using color instance based reasoning. Proposed algorithm extracts target regions with similarity for target color information which is calculated using fuzzy inference for one color value given by human operator. The system shows detected frame with importance rate. Experimental results show that detected frames with the proposed algorithm are 90% of the all goal frame. The algorithm will be used as a basis unit for surveillance camera system.

View full abstract

To deal with a problem of incomplete comparisons in decision maker preference assessment, a SONIA (Self-Organized Network inspired by Immune Algorithm)-based Decision Neural Network (DNN) is proposed. The mutation mechanism of SONIA deals with a limited number of training data resulting from incomplete pair-wise comparisons by decision maker. Numerical experiments on Lp-metric function as underlying decision maker preference show that the error of SONIA-based DNN is 1/4 times lower than that of conventional DNN for decision maker preference assessment with incomplete comparisons. Real-world experiments on dish-up and oven related-work data of a chain restaurant work assignment problem show that the applicability of the proposed method to a real-world restaurant work manager preference modeling.

View full abstract

A correspondence f is called a linear complementarity correspondence if it has a state-variable representation, which can be reformulated as the linear complementarity problem. The paper introduces Classes P and ULT of linear complementarity correspondences, and shows that these classes and the family of all piecewise linear functions coincide with each other. Moreover, the paper discusses the minimum dimensional state-variable representation problem, called the minimal realization problem, and introduces new concepts concerning redundancy in the state-variables.

View full abstract

This paper discusses convergence theorems describing implications between six convergence concepts with respect to non-additive measure: almost everywhere convergence, pseudo-almost everywhere convergence, almost uniform convergence, pseudo-almost uniform convergence, convergence in measure, and convergence pseudo-in measure. The paper shows several new convergence theorems and arranges them together with existing convergence theorems by using ordinality and duality. In addition, it gives a new necessary condition and a new sufficient condition for the Egoroff theorem to hold.

View full abstract

In recent years, in the field of the entertainment robot and the nursing care robot, the method for embedding human emotions to the robot is developed actively. The researches based on not only psychological but also physiological knowledge is also started, and the effect of neuromodulators are gradually making clear in the research field of the brain science. One the other hand, the influence between neuromodulators and meta-parameters of the reinforcement learning system is also reported by some researchers. We have considered a learning method of emotion behaviors of an autonomous mobile robot by using Q-Learning method with the meta-parameters controlled by neuromodulators which is well known in the field of the brain science so far. In this paper, we propose a Q-Learning method with the plural Q-values concerning the maximization and the minimization of rewards and punishments. We aim at the realization of a system that learns complicated emotion behaviors with the behavior selection which has positiveness and negativeness adaptively according to the situation. Furthermore, we also report the result of an experiment by computer simulation to confirm the efficiency of the proposed method.

View full abstract

In this article we show the development and the evaluation of a goalkeeper agent for RoboCup simulated soccer. After we describe the strategies of UvA Base team and its goalkeeper used as our base team, several defects of the goalkeeper are pointed out. We developed a goalkeeper which has a human-like strategy. We show two kinds of experimental results through games with a number of opponent teams. In the first experiment the goalkeeper has the same teammates in its developing environment. This experiment enables us to compare our developed goalkeeper with UvA Base's goalkeeper. The second one involves the combination of the developed goalkeeper and different teams. In this experiment we evaluate the performance of the developed goalkeeper compared to the original goalkeeper from UvA Base and the other team's original goalkeeper. The results of the experiments show the performance of the goalkeeper is improved.

View full abstract

In this paper, we propose a method with dynamic construction facility of the state space, where we initially have no state space, gradually add a new state of fuzzy set and remove unnecessary actions and states.

View full abstract

As Q-Learning is formulated to solve reinforcement learning problems that are modeled as Markov decision processes (MDPs), its learning capability is limited to a task environment whose structure can be dealt with by a single MDP model. For this issue, we propose a new learning method with multiple Q-tables that are flexibly switched over according to the identified dynamics of the task environment. Our learning model consists of two different processing modules subsuming the Q-Learning algorithm: the system identification module and the adaptive state space divider. The former functions as a detector of different environmental dynamics we call "task situation". The latter autonomously regulates the effective resolution of the agent's internal state space by absorbing the distribution of experienced sensory data. This paper explains our basic ideas, and then presents some results from a preliminary experiment using a cart-pole swing-up simulation.

View full abstract

We developed the Remote Monitoring Intelligent System (RMIS) of Local Area Network (LAN) to realize the secure Internet environment. The RMIS has various agents which work to exchange the information related to computer's conditions in the network environment. We shall be able to control the quality and quantity in the network communication by some kind of agents. In this paper, we propose the scheduling method of data transmission by Recurrent Neural Networks(RNNs) to avoid traffic jam in thenetwork. Especially, our proposed RNN enables to expect the good condition at each network device. In order to verify the effectiveness of our proposed method, we report the examination results.

View full abstract

This paper describes a moving object extraction method using a basis function network. The basis function network represents frame images as a linear conjunction of weighted basis functions. The ordinary background subtraction method can't extract the moving object from unstable background. The ordinary frame differencing method can't extract the whole moving object. The proposed method can detect a moving object by the difference between consecutive weights and can reconstruct a local image including a whole moving object. Detections and extractions of moving objects from unstable background are examined and the validity of the proposed method is discussed.

View full abstract

Hardware acceleration of a basis function network which has a good spatial approximation ability is described in this paper. By using pipe-line architecture and simplifying the algorithm, high speed operation (10 times as fast as the previous model) has been achieved. We implement the proposed system to an FPGA and confirm the performance with computer simulations (delay-simulations). By comparing the previous systems, the validity of the proposed system is discussed.

View full abstract

D.Watts and S.Strogatz proposed the two characteristics, L (characteristic path length) and C (clustering coefficient), to define Small World, and they modeled Beta-Graph, which produces Small World Network. This study has triggered lots of researchers in different fields. Recently, it has been suggested that associative memory networks with Small World can provide the same performance as random networks, however these studies have not showed clearly that the networks organize Small World, because their Small World Network is produced by Beta-Graph. Beta-Graph is a simple model which has only one parameter, rewiring probability p, therefore it is difficult to discriminate whether the network constitutes Small World Network or not. To reveal appearance of Small World, it is necessity to measure the networks by using L and C. The purpose of this paper is to organize Hopfield network which constitutes Small World Network and provides the same performance as random networks at small cost, and show that the structure of Small World leads to high performance in associative memory problem by measuring L and C.

View full abstract

In this paper, RBF network structure selection method based on the multi--objective evolutionary algorithms is considered.The candidates of RBF network structure are encoded into the binary coded chromosomes. Then, they evolve toward Pareto--optimal front defined by the several objective functions corresponding to the model accuracy and model complexity. An ensemble network constructed by such Pareto--optimal networks is also addressed in this paper. The properties of the proposed ensemble network is discussed by numerical simulation results in the case of existence of outliers or lack of data, and compared with one selected in the sense of information criteria.

View full abstract

Our class was examined by the questionnaire investigation by students. We divided itinto the level by the scholastic ability and analyzed it. The understanding degree ofthe student's class and a problem for the class were found out.

View full abstract

In this research, a method of fuzzy linear evaluation using theDEA concept is considered. When the evaluation value is calculated by the extension principle, the evaluation values from the linear fnction with the coefficients of fuzzy numbers are given by fuzzy numbers. Here, we express the fuzzy coefficients as the interval of the h-level set, and using the idea of DEA, the evaluation values arecalculated. That is, the most profitable coefficient values in the interval can be selected and the most profitable evaluation value for each object is calculated. Then the evaluation values which reflect the characteristic of the each objects with the maximum concern are obtained by our method.

View full abstract

A fuzzy graph is useful for drawing the relationship in the group and the architecture of the group and allow people to understand that on vision. We show a automatic drawing method of a fuzzy graph with isolated nodes.

View full abstract

We adopted a computer game - the invader computer game - as an example of the interaction between the computer system and the user. We constructed eight different versions of the game. We used Saaty's AHP(Analytic Hierarchy Process) procedure to evaluate how each feature determine whether the game was liked. The results revealed how the features of the game are related with pleasantness of computer games.

View full abstract

We often use a fuzzy graph to analyze inexact information such asthe instruction/cognition sequence, sociometric structure, opinion poll and so on. Concerning the cluster analysis of a fuzzy graph, it is difficult to decide the optimal level of the fuzzy clustering as to a partition tree. In this paper, we would not only discuss a decision method, but also illustrate its practical effectiveness through the case study.

View full abstract

The fuzzy graph is frequently useful for analyzing inexact information such as human relations, community relations and the sociometry analysis. But it is difficult to draw automatically the comprehensive fuzzy graph to analyze them. Actually, it is impossible to process the fuzzy graph with a lot of nodes at short time. So we propose a method to draw a fuzzy graph using the hybrid GA that combines GA with a centroid method. This paper describes the method and a case study to verify it.

View full abstract

Information retrieval system of dish knowledge was hardly formulated in a mathematical model because of complicate relationship among its data items and difficulty in mathematical formula of taste. The authors have designed mathematical model of dish knowledge. The relation among data items are described in predicate logics on the base of relational database model. The relation between taste and seasoning is designed in neural network model. In this paper, the authors discuss the design and implementation of dish knowledge retrieval system on first rank predicate logics and neural network model.

View full abstract

In this school, making the student can touch the piano for the attending school period of two years, and improving to the level by which it can act as a child care person are one of the big problems. The IT society expands rapidly now and the information educational environment has been enhanced also atthe university. The multimedia education was valued, and a diversified approach becomes possible in a plano education. We will report because the current state of the student who entered a school at current year is clarified in the main discourse, and it constructed the environment to support student's piano lesson in "Introduction to the information processing" course.

View full abstract

The authors present an advanced fuzzy auto regression model and show an estimation procedure of their parameters under uncertain conditions.

View full abstract

We propose an identification method of the fuzzy measure using diamond pairwise comparison method.Right and left side of the diamond means ordinal weights' comparison and up and down means interaction degree's comparison.By the comparisons, we estimate a hierarchy structure of the evaluation, interaction degrees of the structure and weights of evaluation items.Using the structure, we identified the fuzzy measure of the comparisons using adjusted lambda fuzzy measure, which is defined to calculate the partial interaction relations.

View full abstract

They can find various types of uncertainty in real decision problems. In the conventional approaches to decision problems under risk probability theory has been introduced to cope with uncertainty. To treat another type of uncertainty this article focuses on possibility and necessity for uncertain information and discus the meaning of possibility and necessity. The decision problems with possibilistic information are formalized and the decision making process is described mathematically. The information processing model is illustrated through two examples.

View full abstract

Independent component analysis (ICA) is an unsupervised technique, which uses higher order statistics than principal component analysis (PCA) to reveal the intrinsic structure of data sets, and is useful for projection pursuit as well. However, in real applications, it is often the case that we fail to extract useful latent variables because they have no connection with predefined criterion variables. This paper proposes an enhanced technique of ICA, which extracts independent components closely related to some external criteria. Preprocessing is performed by using regression-principal component analysis, which estimates latent variables that have high correlation with the external criteria.

View full abstract

In this paper, we propose that the factor for the demand forecast was discovered by using the business control data and meteorological data during year on the assumption of the construction of the fuzzy reasoning model. The analysis of sales of every day and the relation between the number of visitors and the meteorological factor is tried by applying a analysis of variance to these data. Then, the combination of factors with utility by a multiple linear regression analysis is examined from the extracted factor. At this time, to do meteorological data in consideration of the construction of consumer's demand forecasting model with the fuzzy reasoning, and to compose the fuzzy reasoning model easily, the temperature is analyzed rough according to the layer. And we treat it from the result of the analysis concerning the fuzzy reasoning model.

View full abstract

The purpose of this study is to investigate possibilities of applying a schematic representation of fuzzy set formed by concentric circles as an estimation scale to various kinds of measurement. College student subjects were asked to evaluate what degree they are regarded as children and draw an x mark in the suitable position on a concentric circle fuzzy scale. The subjects were also asked to perform some evaluations using concentric circle fuzzy scales. Then the subjects were ordered to evaluate same items with the numbers from 0 to 100. All coefficient correlation between evaluation value using concentric circle fuzzy scales and evaluation value with the numbers from 0 to 100 were significant (p<0.01). Possibility and problems of fuzzy measurements using concentric circle fuzzy scales are discussed.

View full abstract

As the investment model, a chaos short-term forecasting model is used as a forecast technique of stock prices like the option model etc. based on the random walk model who it is a portfolio model who is the diversified investment model, and is the probabilistic model of stock prices, and various models are proposed. However, the debenture allocation model who is the model forecast from the macro data and the market data to judge the investment environment of the claim market price is proposed about the investment in bonds. Here, it proposes a fuzzy model by whom the point of the problem from the business is considered because the feature of the debenture allocation model is discussed.

View full abstract

In this paper, we propose an interior layout support system using interactive evolutionary computation. The conventional system is only concerned with layout of a room, and is not concerned with a usability of layouts.In the proposed system, layouts are evaluated using newly proposed evaluation agents, which behave as virtual residents.As a result, a usefulness value can be calculated as an estimated usability for each layout.Results of experiments show that the layouts that meet the user's demand are obtained.

View full abstract

It is necessary to predict unknown higher-frequency components that are lost by sampling for enlarging digital images. Based on Laplacian pyramid representation, the prediction of unknown higher-frequency components is equivalent to the prediction of a unknown higher-resolution Laplacian image. We have proposed the higher resolution method based on the Laplacian pyramid representation. However, Laplacian pyramid representation was considered for image compression. Thus, we think that the band width of Gaussian filter for image compression is not optimal for digital image enlargement. In this paper, we proposed a new enlargement method based on Laplacian pyramid representation with the variable band width of Gaussian filters. Moreover, we can obtain better enlargement images based on visual evaluation by changing the band width of Gaussian filters.

View full abstract

Multichannel representation of non-linear wavelets is proposed based on an extension of conventional morphological wavelets with 4 channels, in order to perform efficient image compression/reconstruction. The proposed wavelets, especially 3x3 channels wavelets are constructed from min (max) operation in the target window 3x3 pixels and subtractions of each direction pixel and the center of target window that correspond to 1 signal and 8 detailed spaces, respectively. The proposed wavelets can preserve edge information of the original image on the compressed image, therefore, in the case of motion compression, the proposed wavelets are efficient for the compression/reconstruction of P-frames which are correspond to edge images. Through an image compression/reconstruction experiment using standard image database and motion imaginary, it is confirmed the effectiveness of the proposed method compared under the low bit rate compression.

View full abstract