The objective of a fuzzy regression model is to illustrate the potential possibilities inherent in the target system. In the case of the interval fuzzy regression model, the system possibilities are measured by the width of the interval indicated by the model. If the center of the model coincides with the center of the distribution of possibility, more information could be obtained regarding the possibilities of the target system. In the sense that the model should be fit to the human intuition, this would make the proposed model more user-friendly, and increase opportunities for its application. For these reasons we have proposed a fuzzy regression model which has a center that coincides with the center of the possibility distribution. In this paper, population of some district and its number of households as a numerical example, and a basin area and a drift distance in Japanese as an application of the proposed model will be employed to check features of the model.

View full abstract

This paper reports a voice guide reservation system in imaging center for dental clinic. We developed the reservation system by VOISTAGE(R) using the CTI (Computer Telephony Integration) system from NTT DATA Corporation. This made possible for us to accept reservations at any time from pre-registered dental clinics. The reservation system does not require particular environment, and the security is high. The Fuzzy theory was applied in operation. We classified each dentist's seriousness into four stages of "good, normal, bad, and very bad". This classification was based on whether the application form was written properly and correctly, and whether the dentist regularly attended the training which was offered by the imaging center. In each stage, a period which is available to make reservation for dentist is different. A preferential period is offered in 'good' stage and in contrast, postponed period for 'very bad' stage. By this, we expected to prevent unbalanced reservation and an increase in the cancellation.

View full abstract

This paper describes the text localization method from natural images and the performance of the proposed method. The proposed method consists of three procedures, an edge extraction method, a detection of the text location based on the closed edges, and a classification of extracted candidates as text. The distributions of the white and black pixels in text areas are mixed then the proposed information measure is relatively large, whereas the measure is relatively small in the non-text. In the classification procedure, extracted candidates are classified by the information measure. The performance of proposed method is evaluated with the two examples.

View full abstract

It is known that we can evaluate students' work of art, calligraphy etc. objectively by using fuzzy reasoning. However, it is difficult to define how many points one has in each evaluation item. For this reason, we would like to introduce how we used fuzzy number in educational evaluation.

View full abstract

Two types of the growth rate distributions are observed in the analyses of economic data. The growth rate distributions such as profits can be approximated by linear functions on a log–log scale. The growth rate distributions such as sales are difficult to approximate by linear functions because the distributions with a curvature have fat tails. These two types of the distributions have been derived by numerical simulations. By a multiplicative stochastic process, profits types variables have been generated. By superposing the variables not to be negative, sales type variables have been generated.

View full abstract

Visualization of a network helps us a lot to understand structural properties of the network intuitively. Especially molecular dynamics simulation model (MDSM) works well for various types of networks. However, the method has been so far applied to undirected networks. Recently we have develped a MDSM applicable also to directed networks. We demonstrate its usefulness in this talk.

View full abstract

Very often, the position of the current price level is compared to the past average price for price predictions. Most technical analysis utilizes this sort of technical indicators. In the past, daily close prices in the market are used, which has been taken place by tick-wise price data in recent years as intra-day trades by means of online trading system has rapidly become popular. Although the minimum information by means of up/down of the price (binomial) is the most efficient way for the same of rapid decision makings, it is desirable to extend the system to many-valued case, and to include derivatives and the higher derivatives of the price time series. It is the very reason why the novel methodology is called for in which the finer characteristics of price changes by keeping the efficiency of calculation time. I this talk, we report some results of our research on the autonomous decision making system using continuous patterns for the sake to online trades.

View full abstract

This article proposes methods to measure states of market participants based on their behavioral patterns both comprehensively and circumstantially. In order to quantify monopolization of trading patterns the normalized information entropy for relative frequencies of quotations/transactions estimated from high-resolution data of financial markets is employed. The empirical results by using high-resolution data of the foreign exchange market are shown and an agent-based explanatory model is considered in order to understand meaning of proposal indices. It is concluded that the proposal methods may measure participants' behavioral patterns associated with their cognitive patterns.

View full abstract

Mathematical model for hit phenomena is presented. The model is applied to movie entertainment, local events, and social phenomena. The simulation using the model is compared with the blog-entry numbers and incomes for the demonstration of the model applicability.

View full abstract

Companies often carry out questionnaire(s) in order to design marketing strategies or to grasp the trends. Recently, Web questionnaire survey becomes popular with the spread of the Internet in order for companies to reduce cost and to get a lot of questionnaire data. However, in the Web questionnaire survey, some respondents do not answer the questions seriously, because it is not done face-to-face and is done just for giveaway basically. If those answers are included to questionnaire data, there is a possibility that the analysis result of the data is not accurate and shows different characteristics or trends. This paper proposes a quantification of earnestness for answers and a visualization method based on them. It is expected that appropriate analysis can be done by the proposed method.

View full abstract

AHP (Analytic Hierarchy Process) has been widely used in decision making. Inner dependence method AHP is one technique even in case of criteria have dependency. However using original AHP or Inner dependence method, the results often lose reliability because the comparison matrix does not always have sufficient consistency. In these cases, fuzzy representation for weighting criteria is useful. In this paper, we propose fuzzy weights using results from a sensitivity analysis. Through experiment results, we show how inner dependence AHP has fuzziness when the comparison matrix is not sufficiently consistent.

View full abstract

The procedures of scene frame construction in automatic comic drawing from narrative text are studied in this paper. The following two points are considered: (1) Scene frames are generated from verbs expressing character's action, (2) "Sentence importance", "main story importance" and "first importance of appearance of character" are considered for the calculation of importance in a story. Two subject experiments are done for validation of usefulness of the proposed method: (1) comparison of scene frames obtained by the proposed method with those obtained by five subjects and, (2) ten subjects evaluation of scene frames generated by the proposed method.

View full abstract

In this paper we investigate the effect of data partition in constructing fuzzy ensemble classifiers. In the standard version of ensemble methods, the confidence of a base classifier is calculated based on the classification performance of the base classifier on training patterns that are used to construct it. In the proposed method we divide the training data set into two subsets. One subset is used to construct a base classifier and the other is used to calculate its confidence. In the computational experiments in this paper we compare the performance of the proposed method with that of well-known classifiers. The experimental results show that the proposed method is effective for improving the performance of the fuzzy ensemble classifiers.

View full abstract

Under changing environment, which is the environment into which surrounding circumstances change at time, when human operates the equipment, the operators should correctly understand a surrounding environment that changes complexly. And, the operators should judge an appropriate operation within a limited time. The equipment operation support system is necessary to reduce the load of the operator, and to achieve a safe operation. In this thesis, an effective support method of the equipment operation, which presents instruction of operation that user should be going to do in the future, is proposed. In this method, it is important to consider not only present states of surrounding environment but also future states. The way of building expert's knowledge concerning the time change of the states into the computer is proposed, which is "Time change fuzzy sets". The simulation that assumes the lane change of vehicle is done to confirm the effectiveness of the proposed method.

View full abstract

Since the number of rules of the single input rule modules connected fuzzy inference method (SIRMs method) is limited as compared to the conventional fuzzy inference method, inference results gained by the SIRMs method are simple in general. From this reason, Watanabe et al. have proposed a modular fuzzy model which extends the SIRMs method, and shown that this model can obtain good results for reinforcement learning. In this paper, this model is shown to be superior to the conventional SIRMs method by applying to discriminant analysis of a medical data.

View full abstract

The aims of this paper, that predicts the presence or absence and timing of the nasopharyngeal carcinoma (NPC) recurrence by using neuro-fuzzy based techniques. The techniques include an artificial neural network (ANN), adaptive neuro-fuzzy inference systems (ANFIS), the functional-type single input rule modules connected fuzzy inference method (F-SIRMs method) and the functional and neural network type SIRMs method (F-NN-SIRMs method). Five years predictions are carried out. Validity of each predictive model is assured by 10-fold cross validation. The results show that the F-NN-SIRMs method gives prediction performance higher than the other techniques.

View full abstract

Fuzzy modeling is one of the most important techniques for nonlinear modeling. SIRMs has been studied as useful modeling method for actual applications such as control and pattern recognition. In this paper, we study on modeling approach based on Modular Fuzzy Model extending SIRMs architecture. We show construction of the error objective function for modeling of modular fuzzy model and SIRMs affects prediction performance. Through numerical experiments, we discuss about performance of the models.

View full abstract

There is the important relation between the humann memory and consciousness. If the memory were effectively used, there should be the great effecct on the education. We propose the new model of the memory and the effective utilization of the memory.

View full abstract

In climate data analysis for researching how the erath warming is caused, the data has been converted to the sound and stored in the sound recoder iPOD, so that the sound can be repeatedly heard, because the sound was useful to discover the climate categoriea and thier characteristics. Although the datamining is carried out by graphs,the sound prmits us to discover the climate difference between the cities near to the sea and those far from the sea by intuition. The profits of sound application is in the intuitive datamining by hearing. The climate change during the past 50 years in the typical cities of in the categorized climate is shown by graphs.

View full abstract

In the new Couses of Study,The subjects of using or learning Measurement and Control Technology increased.And all junior high school students must learn about the computer measurement control. So we developed Teaching Material of Measurement and Control for supporting Classroom Lessons to high school students or general university students. This teaching Material aims at the presentation of the feedback control such as ON-Off control, PID control and Fuzzy control.And we also propose the systematic method of teaching.

View full abstract

This paper reports on practical examples for algorithm education using physical senses and visual images. It is effective to use physical senses and visual images in algorithm education. Some methods for understanding algorithms are proposed.

View full abstract

We are utilizing the Knoppix system for the education of computer, 2005-2008. In this paper, I introduce the education of the information literacy and the computer programming for the students of junior college of arts.

View full abstract

This paper proposes the situation-dependent membership function estimation method based on analogical reasoning. The proposed method defines the relation between membership functions representing same category in different situations as a function, which is called analogy. If membership functions expressing same categories in different situations are already identified, analogy can be identified by them, and other membership functions in the situation can be estimated by analogical reasoning. This paper also confirms effectiveness of the proposed method experimentally. (1) The relations between membership functions in different situations are expressed by some degree polynomials in preliminary experiments and those polynomials are analyzed to obtain the appropriate degree of polynomials as analogy. (2) Situation-dependent membership functions estimated by analogical reasoning are compared with those identified by other method.

View full abstract

This paper aims at the construction of a friendly system, which plays a seven-card stud poker game with a human player against an opponent player. This paper calls a cooperative friendly system a partner agent. Seven-card stud poker is a kind of a game with imperfect information and it is difficult to consider optimal strategy in such a game. Then, the partner agent has a discussion with a human player on game strategy cooperating with the player. If a human player needs some advice on a game, the partner agent presents its reply. Furthermore, the partner agent presents not only linguistic expressions of strategy but also facial expressions according to the game situation so that the agent has friendliness. Neural network models are used for facial expressions.

View full abstract

This paper presents an imitation learning method, which enables an autonomous robot to extract demonstrator's characteristic motions by observing unsegmented human motions. To achieve this, the robot has to find what he learns from the continuous time series. The proposed learning architecture converts the continuous time series into a discrete time series of letters by using a switching autoregressive model (SARM) after reducing its dimensionality by using a singular vector decomposition (SVD). After the conversion, the proposed method finds characteristic motions by utilizing n-gram statistics referring to minimum description length principle (MDL). An experiment revealed that the proposed method enabled the robot to obtain several prepared characteristic human motions embedded in an unsegmented motion exhibited to the robot. Additionally, its performance is evaluated by comparing that of the proposed method with that of human participants.

View full abstract

This paper considers a multiobjective random fuzzy linear programming problem using possibility and necessity measures. The proposed model is not a well-defined problem due to including randomness and fuzziness, and so introducing chance-constraints and the degree of possibility and necessity with fuzzy goals, the proposed model is transformed into the deterministic equivalent problem. Furthermore, in order to solve the multiobjective programming problem, the model maximizing the minimum aspiration level among all objects is proposed. Then, in order to solve it more efficiently, the relation between absolute deviation and variance is introduced by using the normality of random variables and the efficient solution method based on the linear programming is developed.

View full abstract

A central pattern generator (CPG) and passive dynamic walking (PDW) have attracted much attention in the research field of bipedal locomotion. We describe a motion control method based on dynamic joint passivization for biped robot locomotion. CPG-based motion control is effective for walking on uneven terrain. However, it has serious problems with energy loss. In contrast, PDW saves energy because a robot can walk without any active control or energy input on a downhill slope. However, PDW robot can walk only on a downhill slope, is weak to changes in the environment. We think that active walking needs to be mixed with PDW for robot walking. Our motion control method is based on a mixture of the CPG and PDW, that is, the dynamic passivization of joint control. Experiments using our method successfully generated energy efficient walking and enabled superior gaits.

View full abstract

It is well known that fuzzy mapping plays an important and fundamental role in fuzzy analysis. In this paper, we attempt to determine extensive classes of these fuzzy mappings that are represented by their Taylor expansion. We have first obtained that the continuous function f(x) can transform a fuzzy mapping f(u) using the extension principle. And we discuss some properties of the series of fuzzy numbers. From these properties we have seen that f(u) has the property of being analytic for the analytic function f(x). Finally as applications, we give some examples.

View full abstract

In this paper, we describe an adaptive trading agent which can sell and buy electric power effectively in a locally produced and consumed electric energy network. Recently, to reduce CO2 emissions, and to overcome the problem that fossil fuels will run out one day, we are required to use renewable energies. However, traditional centralized electric network are not suitable to renewable energies. Distributed electric network, such as a micro grid and an ECONET is considered to be an alternative choice. In our model, trading agents manage a storage battery in a minimal cluster, which is a node of local cluster in ECONET. The agent learns a trading strategy by maximizing future cumulative reward based on reinforcement learning method. We build autonomous trading agents based on the natural actor-critic method. Several experiments show that the adaptive trading agents can reduce useless energy consumption and a deficiency in most cases.

View full abstract

This paper proposes a keyword map equipped with a relevance balance controller for supporting data analysis according to compensatory/non-compensatory decision making strategy. A keyword map has been proposed as general-purpose interactive graph visualization tool and applied to various decision making tasks. However, it cannot support data analysis according to compensatory/non-compensatory decision making strategy, because only single relationship type is used. The proposed relevance balance controller enables a user to dynamically adjust weight of each attribute for relationship calculation. Experimental results show the proposed system can provide various viewpoints and support data analysis according to compensatory/non-compensatory decision making strategy.

View full abstract

It is generally thought that living things have desires for conformity as well as desires for differentiation, which make their preferences show trends. Recently, it was confirmed that there were trends in the preferences of how female birds chose their mates. We think trends in the preferences are related to desires for conformity and differentiation and the strength of desires among living species are genetically different from one and another. We describe the strength of desires among living species as being artificial agents of genes. In this paper, we simulate phenomena of fashion in female preferences for a mate by using an agent model that consists of imported conformity and differentiation as genes. In this experiment, we found that there were two kinds of periodic phenomena of fashion and reported the influence of conformity and differentiation on the transition of female preferences.

View full abstract

An image index, extracting DCT domain color and texture features in indexing process, is proposed for JPEG and MPEG fast image retrieval. Feature extraction from 1000 JPEG images requires 2.86 seconds, which is 116.17 times faster than previously reported by Serata (2006) because of efficient compressed domain feature extraction. Retrieval performances of the proposed and Serata's indexes are evaluated on Corel database, and the results suggest that the maximum precision and recall of the proposed index are improved by 22.7% and 11.1% with reduction of indexing times. Due to reduced computational times, the index is applicable to databases which include 100 times larger data sizes than current one.

View full abstract

In this paper, we tried to show covering domain "invisible specialty" of a research fields by offering of the relations between papers in the field to novice researchers by proposed visualization method. We found it is not significantly different between experts and novices concerning the comprehension of actual covering domain of two academic journals in economics by a sorting test of published papers in each journal. Moreover, we confirmed that novices comprehend more clearly and consistently the covering domain of the fields by the developed information visualization system based on hierarchical research field network.

View full abstract