Robot appearance affect the emotions of people who communicate with robots. In order to design a robot which people prefer or love, this study proposes to adopt "Moe (in Japanse)" as a design element. However, "Moe" is a word of ambiguous concept and it isn't simple task to utilize it for robot-appearance design. We built a 3D design system to investigate and clarify the meaning of "Moe" in previous research. As a result, we suggested a possibility that there is a difference in robot appearance designed by whether people understand "Moe" or not. In this paper, we investigate difference of designed appearance in detail.

View full abstract

Recently, the applications of robots, especially in the social service and daily communications, have been widely spread. In order to make these robots and humans live close together, it is important for robots to increase the personal affinity and to have human-like characteristics. In this research, we defined as those that consist of various impressions and personality, and we build robots to have complex and various characters, and we aim at the establishment of method which is adaptable for human characters. In this research, we perform the recognition of body movements of a human using a robot vision, and we build the construction of the system to extract the human character by using the theory of quantification.

View full abstract

Recently, the number of people avoiding communication with others is increasing, and it is considered a growing social problem. According to Johari window, understanding own self through self-analysis and self-disclosure brings confidence in own, and smooth communication. However self-disclosure, talking about own to others, is difficult aspect for some people, because they are afraid of disappointment and denial. In this paper, we propose a network robot agent for talking with people to disclose them. We use robot service network protocol (RSNP) as its protocol. RSNP provides a secure connection, so users are able to talk in a confidential setting.

View full abstract

There are a lot of sorts of ways and devices to telecommunicate in these days. People are able to communicate to anyone anytime anywhere if they would like to. Meanwhile people cann't always perceive cell phone ringing or e-mail arrival. If people feel asleep, their drowsiness may cause them to ignore the notification. If people don't carry or hold a cell phone nearby, they cann't catch the notification. Meanwhile, robots are physical embodiment agents and can move and touch people or objects. Our study adopts the robots' special features in order to assist to let them perceive some notices from telecommunication devices: cellphones, PC and so on. Active Touch communication Robot (AcToR) utilizes touch information instead of visual and auditory information to inform some notifications from telecommunication devices The experiments were performed to confirm whether the touch communication could be valid to inform some notification. The result showed there was a possibility that the touch communication was one of the effective communication tools.

View full abstract

This paper proposes a conversation system to realize natural communication between a human and a robot partner in informationally structure space. This paper consists of three parts. First, we describe the structure of the sensors and robot partner in informationally structured space. Next, we propose a conversation system based on informationally structured space. Furthermore, we propose an utterance learning system based on Fuzzy Spiking Neural Network in order to learn information from informationally structured space and the relationship in the conversation contents. Finally, we show experimental results of the proposed method and discuss the future direction on this research.

View full abstract

In recent years, robots have been active in a dangerous environment such as space and the disaster areas. However, there is a possibility that risk aversion instruction is not in time when a robot became the dangerous scene in such an environment. The robots therefore require acquisition of behavior to avoid danger autonomously. In this paper, we propose a method for avoiding danger using the probability based reinforcement learning (PrRL) to apply to the action acquisition of the robot.

View full abstract

Recently, cooperative operation of a human and the robot through the physical interaction has been studied and developed. As one of the human supporting robots, inverted pendulum two-wheeled mobile robots have been proposed to support human transportation with/without a heavy luggage. In our laboratory, we have studied a suitcase-type two-wheeled inverted pendulum mobile robot that changes the control parameters according to the user intention. The robot recognizes the user intention by the user's physical operation of the robot. It changes the control parameter according to the recognized user intention and supports the luggage transportation. It is necessary to learn the parameters of the control and user-intention recognition user by user because the user preference is different one by one. This paper proposes a method to learn the control parameters and their corresponding intention recognition parameters of the user for a suitcase-type inverted-pendulum mobile robot for appropriate assistive control. In order to show the validity of the proposed method, it shows some experimental results with a real robot.

View full abstract

Recently, huge amount of documents are posted on the web by blog and twitter. These documents are, however, stored in disorder on the internet, thus it is difficult to find required documents. In a lot of documents, tagging is done by hand. However, a lot of documents especially in twitter are not tagged yet. Tagging by managers or authors on these documents is a heavy burden for them, so auto-tagging is needed. This paper tries to construct an auto-tagging system using latent topic information and proposes a new topic model for the auto-tagging.

View full abstract

Eye tracking is a technique whereby an individual's eye movements are measured so that the researcher knows both where a person is looking at any given time and the sequence in which their eyes are shifting from one location to another. In the research field of human-computer interaction (HCI), human intentions can be recognized by analyzing gaze information. On the other hand, people's intentions are largely influenced by a number of objective factors. Therefore, we introduce saliency map as a basis for intention recognition. In the process of saliency map calculation, fuzzy neural network is employed in order to reflect individual characteristics importance of images. In this study, we propose a method realized the intention recognition by combining gaze information and saliency map of images. To verify the effectiveness of proposed method, an interactive teaching system is constructed based on eye tracking and saliency map by using an omnidirectional wheelchair.

View full abstract

In our laboratory, we are developing Interactive Chaotic Amusement System (ICAS) for sound generation using network of chaotic elements. In ICAS, universal sounds can be generated by adjusting pitch, tone length, and volume, which is realized by controlling of synchrony and asynchrony. Several musical elements have been imported in ICAS in order to generate more natural sounds according to different people. At first we focus on the analysis of the relation between the generated sound and 1 /f fluctuation symptom which makes human feel comfortable. According to the results, we found that most of the generated comfortable sounds have the similar characteristic of 1 /f fluctuation. Next, we focus on relaxing people and generating sounds in ICAS using brain wave information. We have developed a system for tuning the parameters of network of chaotic elements to generate relaxation sounds for human. And it can make judgments to the relaxation degree at the same time.

View full abstract

In the research of interactive music generation, we propose a music generation method, that the computer generates the music, under the recognition of human music conductor's gestures. In this research, the generated music is tuned by the recognized gestures for the parameters of the network of chaotic elements in realtime. The proposed sound generation system coordinates with an additional human created main melody. To make outcomes more closer to music, music theories are embedded in the algorithm, as a result, the generated music will be richer. Furthermore, we reconstructed the music generation system and performed the experiment for generating the music composed by human.

View full abstract

This paper proposes an interactive design system of a phrase animation reflecting impression expressed by an adjective. A phrase animation is text expression with motion, color and a melody. The proposed system consists of three parts: the impression estimation part, the motion, color and melody generation and their combination part and the modification part. Experiments are performed in order to confirm whether the proposed system is able to generate phrase animations giving impression of adjective to a generators and appreciators.

View full abstract

In this research, we propose an automatic generation method of melody used interactive genetic algorithm according to specified tone row in order to reflect user's favorite. User provides tone rows to us first, and we design a fitness function of tone rows based on the melodic line which is a part of general music theory and characteristic of provided tones. Next, GA operations such as crossover, mutation and selection are operated in the search space of two octaves, and the groups of tone rows obtained relatively higher evaluation are generated and provided back to user. According to the interactive genetic algorithm user evaluates subjectively these groups of tone rows generated by GA. In this paper, we confirm the effectiveness of this system by constructing the simulator that generates tone rows and accompaniments, and executing the comparison questionnaire of melodies made by the simulator and user.

View full abstract

Recently, remote operation support systems have become more and more important to manage waste incineration plants. It is used to save energy, save manpower and operate safely. Data mining from big data is a key technology to use operator's visual data, noise and vibration of machines, and trend data from sensors for remote operation support. In this paper, we describe the applications of big data to manage the remote operation support system for waste incineration plants. We are now planning to apply big data technology to various fields.

View full abstract

Basal body temperature (BBT) is well known vital index which helps women to estimate ovulation and fertility. In recent years, its network connected applications which propose new manners of BBT utilization have been launched while ICT has been advanced. With the view to such a trend of BBT utilization, authors also developed a new application which gives advices based on their BBT transit and their menstrual cycles to support women's healthcare in their daily life. This paper describes a prediction method of menstruation and ovulation which is built in the developed application. The proposal method predict menstruation and ovulation considering individual variation in menstruation period, hot phase, and basal body temperature transit. In addition, this paper reports analysis results of huge amount of dataset collected via the developed application. According to the results, menstrual cycles have week correlation with age and body mass index (BMI), and basal body temperature has relationship with air temperature. The results have possibilities to improve the proposal predictive method and help to develop new application functions which support women's lifestyle modification for their daily healthcare.

View full abstract

The population of obesity is increasing in not only advanced nations but also in developing ones. Obesity often leads to lifestyle related disease such as diabetes and hypertension. Therefore, it is recommended to improve lifestyles and decrease weight before contracting diseases. However, if people decrease their weight drastically, the risk of rebound weight increases. With the view of this problem, we developed the weight-loss pattern visualization system which provides recommended weight transit based on 443 successful cases. The system supports weight control for 30 days, and the recommended weight transit is converged on individual target weight. In addition, the limit of increase in weight in a day is drawn along with the recommended weight transit. According to our comparative evaluation with a system without weight-loss pattern visualization, the weight reduction of users who used the developed system is greater than that of the comparative one significantly. The results suggest that the weight-loss pattern visualization has more effectiveness on weight control.

View full abstract

Human beings are living in the dynamic environment where the surrounding situation is changing every moment. In this environment, when they determine action depending on the surrounding situation, we use the knowledge in consideration of the change of state of space and time. There is right turn operation at a crossing as an example using this knowledge. At a crossing, the crossing person of an oncoming car or a pedestrian crossing from whom the surrounding situation changes with time, and human beings recognize these motions immediately and determine right turn operation. Thus there are many operation beginners' accidents and many drivers who are not good at performing right turn operation. Therefore the purpose is that I propose space-time fuzzy inference in order to include knowledge including the change of state of space and time in a computer, it applies to right turn operation at a crossing. In this paper, it explains details of the proposed space-time fuzzy inference, and the effectiveness is evaluated by the experiment.

View full abstract

What people look at when they stroll through a street? This is an important problem for tourism industry. Presumably, what the visitors look at is different from what the residents do. The authors conducted an experiment in which the residents and the visitors strolled through a street and took pictures. As a result of the quantitative analysis, it is Shown that the residents paid their attention at the height of eye level whereas the visitors allocate their attention at various height. It is because the important information exists at the height of eye level for residents, on the other hand, the visitors intend to under their surroundings.

View full abstract

Japanese ICT companies still have been struggling with poor performance in spite of various management reforms since collapse of the bubble economy in the early 1990s. Therefore, we recognize the necessity to understand issues of their technology innovation strategies. Specifically, this paper discloses and analyzes patent strategies of Japanese and global large ICT companies, Sony, Sharp, Panasonic, Samsung and Apple in smart phone market, whose share has increased in ICT industry.

View full abstract

Rough set theory was proposed by Z.Pawlak in 1982. This theory can mine knowledge granules 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 apply the decision rule to reason, estimate, evaluate, or forecast an unknown object. In this paper, the rough set theory is used to analysis of time series data. Knowledge granules are minded from the data set of tick-wise price fluctuations. We search for the law of similarity from time-series data using the rough sets.

View full abstract

At 1978, the value of production of the food industry reached 25% of GDP in China, and it developed into the biggest Chinese industry. After that, the food industry in China also maintains high growth until now. The development of the food industry in China has converted and developed into the model of the major company formed as an industrial group from the direction of small and medium-sized enterprises especially. Moreover, the maturity of the correlative industry in China is also increasing, profits came to concentrate on a major company now, and it makes much brand of famous food appeared as Chinese industrial technique improved. As the economy and population continuously grow up of China, the grade amount of the demanded and pluralization is progressing. Therefore, it is expected that a big development, such as the huge market of the food industry in China. In this paper, we analyze the external environment about the huge food market in China. And we consider a management strategy from the viewpoint of the major company in such intensely competitive market.

View full abstract

In 2012, 15.13% of the total fiscal medical care expenditure was for lifestyle-related health care costs, which was approximately 179 billion yen. Lifestyle-related diseases are not only the biggest factor in reducing healthy life expectancy but also have the most significant impact on the national medical care expenditure. In addition, lifestyle-related diseases can be prevented by moderate daily exercise, a well-balanced diet, and not smoking.Our fuzzy robust regression model is a controllable model describing a target system. Therefore, our model is used to analyze the relation between medical care expenditure and selected lifestyle factors.

View full abstract

Transportation Problem (TP) is a well-known basic network model that can be defined as a problem to minimize the total delivery cost. But for some real-world applications, the TP model is often extended to satisfy other additional constraints or is performed in several stages. In addition, traditional TP model is not include the concept of opportunity loss. In traditional TP, customer demand is always satisfied. In the real world, there are many cases where customer demand is not satisfied, and opportunity loss occurs. In this paper, we formulate a two-stage transportation problem with opportunity loss. To solve the problem, we applied Differential Evolution (DE).

View full abstract

Focusing that the equation of the exponential smoothing method (ESM) is equivalent to (1,1) order ARMA model equation, a new method of estimation of the smoothing constant in the exponential smoothing method was proposed before by us which satisfied the minimum variance of forecasting error. In this paper, we utilize the above stated theoretical solution. Firstly, we estimated the ARMA model parameter and then estimate the smoothing constants. Thus the theoretical solution is derived in a simple way and it may be utilized in various fields. This new method shows that it is useful for the time series that has various trend characteristics. The effectiveness of this method should be examined in various cases. Keywords: minimum variance, exponential smoothing method, forecasting, trend, genetic algorithm

View full abstract

In recent years, the needs for intermittent demand forecasting are increasing because of the constraints of strict Supply Chain Management. How to improve the forecasting accuracy is an important issue. There are many researches made on this. But there are rooms for improvement. In this paper, a new method for cumulative forecasting method is proposed. The data is cumulated and to this cumulated time series, the new method is applied to improve the forecasting accuracy. The forecasting result is compared with those of the non-cumulative forecasting method. The new method shows that it is useful for the forecasting of intermittent demand data. The effectiveness of this method should be examined in various cases.

View full abstract

In industry, making a correct forecasting is inevitable. If the correct forecasting is not executed, there arise a lot of stocks and/or it also causes lack of goods. Time series analysis methods are applied to this problem. In this paper, a hybrid method is introduced and plural methods are compared. Focusing that the equation of exponential smoothing method(ESM) is equivalent to (1,1) order ARMA model equation, a new method of estimation of smoothing constant in exponential smoothing method is proposed before by us which satisfies minimum variance of forecasting error. Generally, smoothing constant is selected arbitrarily. But in this paper, we utilize the above stated theoretical solution. Firstly, we make estimation of ARMA model parameter and then estimate the smoothing constant. Thus theoretical solution is derived in a simple way and it may be utilized in various fields. Furthermore, combining the trend removing method with this method, we aim to improve forecasting accuracy. An approach to this method is executed in the following method. Trend removing by the combination of linear and 2nd order non-linear function and 3rd order non-linear function is executed to the original production data of two kinds of confectionery. Genetic Algorithm is utilized to search the optimal weight for the weighting parameters of linear and non-linear function. For the comparison, monthly trend is removed after that. Theoretical solution of smoothing constant of ESM is calculated for both of the monthly trend removing data and the non- monthly trend removing data. Then forecasting is executed on these data. The new method shows that it is useful for the time series that has various trend characteristics and has rather strong seasonal trend.

View full abstract

In industries, how to improve forecasting accuracy such as sales, shipping is an important issue.There are many researches made on this. In this paper, a hybrid method is introduced and plural methods are compared.Focusing that the equation of exponential smoothing method(ESM) is equivalent to (1,1) order ARMA model equation, new method of estimation of smoothing constant in exponential smoothing method is proposed before by us which satisfies minimum variance of forecasting error.Generally, smoothing constant is selected arbitrarily. But in this paper, we utilize above stated theoretical solution.Firstly, we make estimation of ARMA model parameter and then estimate smoothing constants.Thus theoretical solution is derived in a simple way and it may be utilized in various fields. Furthermore, combining the trend removing method with this method, we aim to improve forecasting accuracy.An approach to this method is executed in the following method. Trend removing by the combination of linear and 2nd order non-linear function and 3rd order non-linear function is executed to the manufacturer's data of sanitary materials.The weights for these functions are set 0.5 for two patterns at first and then varied by 0.01 increment for three patterns and optimal weights are searched.Genetic Algorithm is utilized to search the optimal weight for the weighting parameters of linear and non-linear function. For the comparison, monthly trend is removed after that. Theoretical solution of smoothing constant of ESM is calculated for both of the monthly trend removing data and the non-monthly trend removing data. Then forecasting is executed on these data. The new method shows that it is useful for the time series that has various trend characteristics and has rather strong seasonal trend.

View full abstract

In Japan, the number of labor accident in the construction industry shows still high value as ever. According to the statistics of last year, the number of labor death accident in the construction site shows about 400, and this number holds one third of the number of labor death accident in all industry. It was also made clear by the statistics that this number is the most in all industry. It is the present situation that time and cost for the safety education in the construction site is decreasing as the result of severe price competition. In spite of having taken a lecture on safety education, many problems of causing the accident which was originally able to be prevented have also been reported because of the too self-consciousness. In this research, we paid our attention between the concentration to the job and the pulse of the worker, and we developed the reminder system to the unsafe behaviors by using the sensing data of worker's pulse.

View full abstract

The discussion is required in actual traffic safety countermeasure planning with empirical knowledge of experts and the field survey results. The decision of experts are recommended according to many findings based on the facts with fuzzy information. Therefore, fuzzy reasoning model for decision making is proposed in the study to classify the cause of traffic accidents in intersections, direct the traffic safety countermeasures and propose the individual traffic safety countermeasures. Since the real examples of traffic safety countermeasures are observed, the proper formulation of intelligent information processing with fuzzy reasoning can be proposed.

View full abstract

Recently, Hybrid vehicles (HV) and electric vehicles (EV) are regarded as clean energy vehicles. In future low carbon city, the carbon dioxide emission should be reduced in urban transport section. Therefore, the urban transport system is regarded as a major issue in the smart city for future environmental sustainability. The study aims at investigating the introduction of clean energy vehicles (CEV) of trip makers in increase number of several low carbon vehicles. In the study, the vehicle ownership can be formulated respectively with fuzzy logic. The decision model of trip maker can be created to change the vehicles from petrol vehicle to electric vehicles as well as hybrid vehicles. Therefore, the demand estimation model for advanced type vehicles can be developed in terms of addition of travel mode types in the low carbon society. The model should be designed to formulate the singleton fuzzy reasoning in vehicle ownership of trip makers. The models may demonstrate the essential elements in traffic demand estimation of future smart cities. Finally, the results of the study can be effectively utilized to create the multi-agent models to describe the spatial distribution of hybrid vehicles and electric vehicles to reduce the carbon dioxide emission in the city.

View full abstract

In the local city, public transport users decreases by the motorization. On the other hand, with the progression of aging, the increase in elderly person having difficulty in driving of the vehicle is predicted. Community bus is operated by the local government to secure the mobility in the local city. However, the community bus has some problems such as little ridership. In the study, the estimation model for travel demand is developed by using the questionnaire survey data to analyze the consciousness structure. Users of the community bus are few among questionnaire respondents. In the study, the rough sets is used to build the model. Therefore, the development of a model having high explanation is possible.

View full abstract

Multi-agent simulation approach is suitable to describe the heterogeneity on social interaction. The choice of clean energy vehicle corresponding to the transport policy related to economic incentive is estimated with the multi-agent simulation. The proposed multi-agent simulation system consists of the vehicle choice model and the social interaction model. The vehicle choice is affected by the concern for environment and the social conformity effect in the model. The relation between agents like a social network of the real world is generated with a simple algorithm by small world network model. The time series changes of the number of the clean energy vehicle and the volume of greenhouse gas emission are estimated by proposed multi-agent simulation.

View full abstract

Image with a surface distortion reduce accuracy of matching and OCR. Able to reproduce the plane image is enabled to measures of reduced accuracy. Therefore we propose the method to estimate the target cross-sectional shape, and to reproduce planar surface.

View full abstract

Image registration (IR) has been used in brain function analysis, voxel-based-morphometry, and so on. The conventional IR methods in MR images mainly use MR signal based likelihood. However, they cannot prevent miss-registration to different gyri because they do not evaluate correspondence of sulci. Also, we cannot directly apply methods for adult brain to neonatal brain because there are large differences in MR signal and sulcal width. This paper proposes a non-rigid 3D IR method using flattening with sulcal-distribution index (SDI) which is calculated from MR signal around the cerebral surface. And, control points are spatially smoothed by using a spring model. The likelihood used is mutual information of SDI. The method will be effective for neonatal brain in which the accurate delineation of cerebral surface is difficult. Results in 7 neonates (modified age; 3 weeks-10 months) showed that an angle difference between a landmark was improved about 20% in comparison with the conventional method.

View full abstract

Kinect is a sensor device which has been released by Microsoft. Kinect has function of tracking the skeleton informations of human being. However, in the crowd, like the event site environment, mistracking may occur. In other words, Kinect loses the tracking ID of the target person when the sensor is interrupt by any other object or person. Then, the target person is re-allocated a new tracking ID, Making the previous tracking ID is invalid. However, Kinect doesn't recognize the new tracking ID and the old one to be of the same person. Maybe, Kinect tracking the another person. Therefore, in this paper, we propose the method for a new and precise tracking ID of the TARGET person by holding the informations of the position even when the target disappears from the Kinect sensor. Besides, we confirmed effectivity of our method through same experiments.

View full abstract

Chain Form Reinforcement Learning (CFRL) was proposed for a reinforcement learning agent using low memory. However, we hold unused information in the memory. In this paper, in order to make memory small dramatically, we introduce Turning Spot Learning (TSL). TSL is a method which imitates human perceptions. If we are asked direction, we often tell a suitable turning spot. Based on this, a TSL agent learns only suitable turning spots. In order to select an action, a TSL agent uses distance (the number of actions). Our method was made a comparison to Q-Learning and CFRL in two kinds of goal search problem. We examined performance and discussed the best usage environment.

View full abstract

UCB algorithm was proposed as one of the action choice methods used in a multi-armed bandit problem. In this method, an agent chooses the action by comparing upper bound of confidence intervals of estimated values, thereby it has a better performance than others, like ε-greedy. In this paper, we proposed the method to apply UCB algorithm to Q-learning, and experimentally evaluated its performance by the shortest path problem in the continuous state spaces.

View full abstract

Reinforcement learning is a method to learn the optimal behavior through trial and error in an unknown environment. If the environment is strongly non-stationary, the agent takes a long time to learn the optimal behavior. There have been various studies in order to solve this problem. As far as we know, these methods have structure which consists of recognition of environmental change and response to environment. In the conventional method, agent has sensor to cognition environmental change and switch the optimal behavior and the exploring behavior. In our method, the optimal behavior and the exploring behavior can be decided according to probability distribution by Bayesian updating state values of probability distribution.

View full abstract

Research on the on-line learning for fuzzy classifiers has been conducted without investigating the treatment of previous training patterns. This research investigate the influence of the previously available training patterns on the classification performance for the on-line learning where training patterns are given in a streaming manner.

View full abstract

The purpose of this paper is to reconsider the given comparisons, since they never represent the decision maker's judgments accurately. His/her judgment of two items may not be an exact value and s/he may not fix an exact wight of each item. In the conventional Interval AHP, reflecting the weight uncertainty, the weight of an item is denoted as an interval. This paper assumes that the weight of one of the items is certain in order to consider the comparison uncertainty. In other words, the item is assumed to be a standard in comparing the items. Then, a comparison can be estimated from the interval weights obtained by assuming one of two corresponding items as a standard. There are two estimations of the comparison and the interval which is included in both estimations is reliable so that the given comparison is replaced into the interval.

View full abstract

GRIP method is a useful tool for narrowing down the alternatives by all addtive utility functions compatible with given pairwise comparison data in multiple criteria decision making problems. It has been utilized pairwise comparison data showing which alternative is preferred and which pair of alternatives has a bigger preference difference. In this study, we propose to utilize the ratio data of preference intensity difference between alternative pairs and the ratio data of criterion-wise preference intensity difference between alternatives. We demonstrate that we preserve the tractability of the method and a more informative robust preference relation is estimated by the utilization of those ratio data.

View full abstract

Extended SpikeProp, which is a kind of spiking neural networks, can learn the timing of spikes. Our research group has discussed the method to learn not only the spike timings but also the number of spikes on extended SpikeProp. But, the proposed method is sensitive to the number of hidden units. In this article, we discuss the method to adjust the number of hidden units. Our method improved the rate of successful learning from 80% to 95% in simple experiments.

View full abstract

Discretization of continuous attributes is a key issue in classifier design from numerical data. In the fuzzy systems community, generating membership functions from numerical data has been an important research topic. Fuzzy genetics-based machine learning (GBML), which is one of the frequently-used techniques to design fuzzy rule-based classifiers, has often used uniform fuzzy partitions to generate initial membership functions. In this paper, we apply a class entropy measure to the discritization of numerical attiributes in order to generate inhomogeneous interval partitions. Nonuniform asymmetric membership functions are constructed from the generated interval partitions by introducing a parameter called "fuzzification grade" (0: interval, 1: full fuzzification). Through computational experiments, we examine the effects of the fuzzification grade on the accuracy of fuzzy rule-based classifiers. We also propose an ensemble classifier which has several fuzzy rule-based classifiers with different fuzzification grades.

View full abstract

The purpose of this research is to investigate the relation between the present self-recognition and future self-recognition in young, middle-age and elderly groups with the multi-axis concentric circle scale. The participants were asked to evaluate proximity between themselves at present and themselves at age from 0 to 100 with the multi-axis concentric circle scale. Then the participants were also asked to evaluate the degree concerning satisfaction with life, happiness, confidence in health, hopefulness and volition that they had about themselves at age from 0 to 100 with the circle scale. The participants were also asked to estimate nine characteristics at present with numerical values (0 to 100). The evaluation scores of the proximity item were affected according to the difference between a participant's present age and the age of 90. Therefore, in order to compare the difference in correlation among the three groups appropriately, the evaluation scores of the proximity item were not included in indices of future self-recognition.The results showed that the correlations between the mean scores of the numerically evaluated nine items about the present self-recognition of each participant and the mean scores of the five items about the future self-recognition of each participant on the circle scale were statistically significant (p<0.01) in the elderly group. However, the correlations between the same items were not significant in the young and the middle-age group. The possible factors (difference in age, birth cohorts, etc.) of the difference in the recognition about the life-span-development process between the two groups were discussed.

View full abstract

Recently, the number of elderly people with dementia has been increasing, and this trend would be a serious social (and medical) problem in Japan. Thus, preventing and improving of dementia are very important. To solve the problem, many researches on robot-assisted therapies have been reported. Currently, various check tests are also conducted to evaluate progression of his/her dementia symptom in many welfare facilities. Some elderly persons are, however, often so nervous about the tests and consequently, these cannot be did with appropriately. These tests have to be conducted without their awareness of the tests. To solve the problem, we have been developing a system using daily conversations with a conventional robot to evaluate progression of dementia without his/her awareness. However, the conventional system could evaluate only time/geographical orientation and short-term memory. In this paper, the authors developed a new dementia evaluation system using drawing tests with a touch panel display as well as that using daily conversations.

View full abstract

With development of the Internet, media lessons has also developed greatly. Recently, student's needs is increasing for media lessons, we have to satisfy student's needs. In this paper, we propose a needs analysis applying Kano model.

View full abstract

In this paper, the authors discuss structure analysis of instruction items. We describe how to get the similarity structure graph, the partition tree, the connectivity structure graph, the approximate structure graph, and the cognition structure graph after scoring the quizzes in the classroom. We also discuss the case study concerning exponent and logarithm in high school mathematics education.

View full abstract

Toward the realization of effective distance education, a preliminary system with atmosphere information feedback is proposed based on fuzzy inference using customized knowledge for learners' interaction. It aims to realize a new learning experience such as attendance to a lecture in a virtual classroom, and provides information, which is necessary to revise the contents, to system manager. Evaluation, called POMS test, of learners stress is carried out for 7 graduate school students after finishing the distance education with a CAI web lecture in which their atmosphere information in a virtual classroom is illustrated on the lecture screen. The learners’ stress measured by POMS test in proposal is decreased (T-A, D, A-H, V, F, C) = (-4.3, -0.14, -1.3, -0.14, 0.29, -0.43) by compared with one of traditional distance education.

View full abstract

A new region segmentation and conspicuous object extraction method is constructed based on virtual super pixel and global features. The simulation results are presented. The conspicuous object region was obtained through the following process: (1) extraction of the background regions, (2) calculation of the conspicuity of the segmented region, (3) unification of the low-conspicuity regions to the high ones. The conspicuous objects extracted through the proposed method matched those extracted by humans. Proposed method may be a model of a middle information representation in the brain.

View full abstract

The present study considers a fuzzy color system in which three membership functions are constructed on a color triangle. For a given fuzzy input, this system outputs the center of gravity of three weights that are associated with respective grades. Three fuzzy sets (red , green , and blue) are applied to the color triangle relationship. By evaluating the attributes of redness, greenness, and blueness, a target color can be easily obtained as the center of gravity of the resulting fuzzy set. The output of the system is a tone triangle, which includes a compound vector with three weights (scalars) in color space. The difference between a fuzzy input and the resulting inference output is shown by the input-output characteristic ( leg shape) between the redness of the input and the chromaticness of the output. Shape effects of conical fuzzy set are investigated.

View full abstract