We are developing an indoor movement support machine to conduct livelihood support for people with lower limbs disabilities. A joystick is designed for users to manipulate this machine. However, it is inconvenient for the user to operate the joystick and the same time conduce daily life activity. Therefore, we are aiming to develop a method to detect the movement intention by the movement of user's upper body. In this report, we measure the specific movement by inertial sensor and identify the movement intention using a hierarchical neural network identification method. The effectiveness of the identification method is discussed.

View full abstract

In order to synchronize the movement of omni-directional walking training robot with the walking motion of users, the walking training need to recognize the walking motion of user and move following the user's waking motion. The use's walking motion could be analyzed and identified based on the surface electromyogram(sEMG). In this report, the sEMG of muscles which were utilized in walking motion was measured. A hierarchical neural network was used to identify the walking motion. The usefulness of the identification result was discussed by experiment.

View full abstract

Use of a myoelectric prosthetic hand is an effective way to rehabilitate upper limb motor function of amputee. In our previous study, we have already developed a two DoF prosthetic hand which has driving MP (meta phalangeal) joint of interlocked middle finger and index finger, and CM (carpometacarpal) joint of thumb. It is structured to perform three grasping postures, namely, power grip, precision grip and lateral pinch. To realize holding function with these postures, 85% of hand movement for ADL is covered. Though the grip performance of the hand depends on joint angles of thumb and fingers, the shape of hand was decided on an empirically. Therefore, the problem that we cannot design the shape of hand according to desired function was remained. It should be solved to clarify the relationship between hand functions and shape with a proper shape model. In this study, we proposed a shape model for function design of the two-DoF prosthetic hand, and confirmed the hand function by some experiments. Then, we measured their gripping force by pull-out test, and evaluated the hand function in daily life by Pick-and-Place task with selected target which are frequently used in daily life. The result shows following tendency: (1) the bending angles of fingers and thumb are larger, their pull-out force are larger, and their grasping function- are lower, (2) the bending angles of fingers and thumb are smaller, their pull-out force are smaller, and their grasping functions are higher.

View full abstract

In this paper, in order to perform a walking neuro-rehabilitation of brain cognitive function, we developed a system that could analyze the brain activity during virtual walking and feedback the analysis result to subjects. The important thing in neuro-rehabilitation by virtual walking is that subjects should be informed with imagination situation to recall the sensation of walking. In this report, we measure the brain activity by a optical topography device when image presentation was feedbacked to subjects and the effects of feedback information on virtual walking was discovered.

View full abstract

The purpose of this study is to support an independent life of person with the acquired visual impairment. In this study, an attempt is made to develop a walking support system comfortable for acquired visually impaired users. Support systems for a person who has the congenital visual impairment have been developed by various researchers. In existing researches, walking support systems have been used by combining with the Braille and the white cane. However, the acquired impaired feels stress against the use of these systems. This is because the life with the visual impairment is extremely-different from the past custom. In addition, the treatment as the visually impaired injures that person's pride. Therefore, it is necessary to develop a support system that considers a person with the acquired visual impairment. In the proposed system, an attempt is made to support a walking with a small wearable camera in consideration of the comfort of user. In this paper, several experiments are presented to demonstrate that the proposed system can detect a person who has possibilities to collide with a user.

View full abstract

In this research, we developed the simulator that can make experience of the localized severe rain for raising the ability in sensing for the risk. The effectiveness of this simulator was evaluated by comparing with the disaster prevention education based on the past disaster documents such as the damage number, the damage photo, the radar rainfall information and so on. Furthermore, in order to verify the effect, in disaster prevention training Takatsuki Akutagawa district, get to experience actually this simulator to training participants, we examined the effect of auditory information and visual information is given to the crisis prescient.

View full abstract

It is important task to analyze a damage or dangerous situation after a disease and prepare the emergency action for disaster mitigation. In this paper we propose the evacuation simulation system after natural disaster using multi-agent system. In the proposed simulation system human elements are represented by dynamic model, and the evacuation flows in some configurations are visualized. We discuss the performance of the proposed model to represent the emergency situation.

View full abstract

The purpose of this research is re-implementation of fuzzy library with programming language Go and extending for parallel programming. Go is new system programming language. Furthermore, an excitable binary file like C language can output, and a parallel programming and the garbage collection are supported. The research of 1995 years defined Fuzzy Description Language(FDL) which provide simple syntax of fuzzy inference. After it is translated to C program code which call the fuzzy library, it is compiled to a executable binary file. But this fuzzy library not enough supports recent programming. Recently computers speed up by multi-core CPU, therefor fuzzy library should support parallel programming. Thus we implement the fuzzy library in Go.

View full abstract

An editor system on tablet pcs for multi dimensinal fuzzy set; Konohen fuzzy set is proposed. It is very difficult to touch a multi dimensional fuzzy set that is de ned on three or more dimension space. We show an new system that use muti touch property of tablet pc combined with geometrical structure extraction technique to edit Konohen fuzzy set.

View full abstract

By inputting a Keyword its related Information spreading on Web can be easily collected. The users often read the text by skim reading in front of the vast information. But we think that it may be possible that the skipped information is useful, which we cannot find only by developing higher level Recommender Systems. In these years we concider not only the Recommender Systems accuracy but also the evaluations and proposals by Serendipity metrics have been studied. There have been many metrics proposed by the researchers, but we don't have any standardized one yet. In this research, by using two metrics, the already-developed Recommender System, we examine if it recommended Serendipity information to the users: one is Yuan’s serendipity metric and another is Adamopolous' serendipity metric. Yuan's metric showed a high value 0.9055, instead Adamopolous’ metric showed a low value 0.1262. Adamopolous metric seeks the quantity included in all recommended lists. The Serendipity information does not appear frequently, so we expect the values are inevitably low. Accordingly I think we should create an metric by considering and understanding that the Serendipity information is not the information which often appears.

View full abstract

Recently, with the spread of wearable devices and digital cameras, the opportunity has increased to record everyday things in multimedia data, such as sound, photo or video. For users who record these as huge life-log data, it is not easy to search the data what and when the user wants. Under such circumstances, the users usually try to find them with any meta-information like recording date, the location or people within, for example. However, if these meta-information are missing or no one tagged it, the users have little chance to reach them. In this paper, we propose an image retrieval system like a WWW search engine. In proposed system, users will put the text query to the search engine as usual. The specific data in their private database are searched with the similarity of the data deployed by the search engine.

View full abstract

In recent years, intuitive style operations and the user-friendly operations are necessary for the future computer interaction. HMD and camera devices are usually used on those style systems and research. For beginner operators, HMD is easy devices at anywhere and anytime. But there are some problems. Especially, the hand detection is difficult. The hand area has specific dynamic colors. It is not robust for illumination change and individual difference of flesh color. For those problems, our research selects the color distribution of hand area for the gesture operations. Our method renews color distributions dynamically about hand areas by the characteristic tracking. This method has robustness at illumination change and individual difference of flesh color. In addition, this method runs only processing of tracking area. It cuts calculation cost. The implementing the proposal method has an advantage at real time.

View full abstract

Word selection has a large influence on communication of feelings or sensations in the daily discourse necessary to deepen our interpersonal relations. In the Japanese language, onomatopoeia-like words play an important role in communication sensual impressions. However, not all such terms are sound oriented. In this paper, we describe the methods and results of a neural network model for inferring the relation between a given phoneme pattern and the degree to which the word evokes an aural sensation. The objective of this research is to build an interface that enriches our sensory experience in human-computer interaction by modeling the synesthetic relation between phonemes and naturally occurring sounds.

View full abstract

Differential evolution is a population-based stochastic search technique for solving optimization problems in a continuous space. DE belongs to the class of evolutionary algorithms and is similar to Genetic Algorithms (GA). Due to its simplicity, effectiveness and robustness, DE has been applied to a variety of real-world problems. Interactive Differential Evolution (IDE) is a kind of Interactive Evolutionary Computation (IEC) that user evaluate the individuals for generation alternation. In this paper, we propose a fashion coordination system that supports users to choose their favorite color coordination. In IEC, it is important to reduce the fatigue of a user during the evaluation process. Therefore, we introduce evolutionary level to DE algorithm and propose an fast method for reducing the number of evaluations. Finally, the proposed system confirmed the validity through experiments.

View full abstract

In regional community, there are exist various regional problems, and such as local government, community people, business establishment and NPO are deal with these problems with partnership. In this study, in order to support approach to regional problems in regional community, it is intended to visualize the regional condition by analyzing the data properly. In this paper, I consider the easy-to-understand presentation of regional data using residents questionnaire data about image of the river and time-series data of population of the region.

View full abstract

In recent years, 3D depth sensor became popular and we can purchase them at low price. However, a large amount of 3D data complicate real-time processing. In order to improve the processing speed for 3D depth data, we propose a data structure, called Multilayer Mosaic. In this paper, we developed a system that detects surfaces and calculates their angles from depth data using Multilayer Mosaic. The experiment shows the improvement of processing speed without losing much accuracy.

View full abstract

The focus of fuzzy c-means (FCM) clustering method is normally used on numerical data. However, most data existing in databases are both categorical and numerical. To date, clustering methods have been developed to analyze only complete data. Although we, sometimes, encounter data sets that contain one or more missing feature values (incomplete data) in data intensive classification systems, traditional clustering methods cannot be used for such data. Thus, we study this theme and discuss clustering methods that can handle mixed numerical and categorical incomplete data. In this paper, we propose some algorithms that use the missing categorical data imputation method and distances between numerical data that contain missing values. Finally, we show through a real data experiment that our proposed method is more effective than without imputation, when missing ratio becomes higher.

View full abstract

In Exploratory Search, repetition search is required. When the retrieving person spends much time on selection of pages, the efficiency of search remarkably falls. I think that the efficiency of Exploratory Search improves by classifying a page based on similarity. Since the WEB page cannot classify a class alternatively, I use Fuzzy Clustering. This technique clusters the morphological-analysis result of the snippet of search results based on the vector given by the evaluation function. The classification result by fuzzy clusteri ng is shown and application to Exploratory Search of fuzzy clustering is considered.

View full abstract

Fuzzy c-means clustering using entropy maximization method consists of an entropy function, an annealing method, a cooling schedule, and normalization constraint. There exist a few possibilities for each component, for example, Shannon entropy, fuzzy entropy and so on. An optimal combination and an optimal setting of parameters of these components for a given dataset are still unknown. In this article, by changing the combination of the components, we examine and clarify characteristic of each method from this point of view.

View full abstract

The important thing in Exploratory Search is that a retrieving person clarifies the goal of search. For that purpose, it is required to find the keyword which related to Search-word and pulls a retrieving person's interest. However, since the information acquired by search is huge, it is difficult to find the keyword which fulfills conditions from the information. Then, I thought that such a problem was solvable by developing the tool which extracts only required information from search results and displays the clustered result. In order to make a clustering result intelligible visually, a self-organization map is used, and information is arranged and displayed on a two-dimensional map. Moreover, in order to be able to reflect a user's idea in a clustering result, it enables it to change freely the parameter of the feature vector used by SOM. Finally, evaluating the usefulness of this tool by experiment.

View full abstract

This paper proposes a new image representation by the possibilistic fuzzy clustering about the object classification. In the image representation, the k-means algorithms is generally used. Our proposal technique is used the possibilistic fuzzy clustering, and adjusts membership width of clusters on the feature distribution. In addition, our technique is based on VLAD image representation. Finally, we compare our technique to recent technique using Caltech101 dataset.

View full abstract

This paper firstly discusses two fuzzy inference models with interval input and fuzzy input. Second, a fast calculation method of the fuzzy inference models is proposed. Third, it clarifies properties of proposed fuzzy inference models. Finally, it shows that the proposed model can be easily obtained from equivalence property.

View full abstract

It is well known that fuzzy numbers plays an important and fundamental role in fuzzy analysis. In this paper, we attempt to define VSOP fuzzy number using an ordered pair (f, g). We have first discussed the addition and scalar multiplication for VSOP fuzzy numbers. From these operations we obtain some properties of VSOP fuzzy numbers.

View full abstract

This paper gives another proof of Sugeno's theorem [Theorem 1, Fuzzy Sets and Systems 222 (2013) 1–17] to calculate the Choquet integral of monotone functions on the real line. Sugeno’s original theorem assumes that the integrand is continuous. However the presented new proof shows that the conclusion of the theorem is satisfied without the assumption of continuity.

View full abstract

In this paper, we have introduced Zadeh's evaluation methods for natural language propositions involving fuzzy qualifiers (possibility, probability). We present research plans for these themes, for evaluation methods for fuzzy quantified propositions, and for inference methods for the fuzzy quantified propositions, in the future.

View full abstract

In this paper, I report that, by studying the history of logic algebras, I found fuzziology or fuzzy logic is in the main stream of logics and mathematics.

View full abstract

Deep level emotion understanding and its visualization method are proposed for multi-agent communication such as humans-robots communication, where customized learning knowledge of an observed agent is used with the observed input information from Kinect. It aims to realize agent dependent emotion understanding by utilizing special customized knowledge of the agent, rather than ordinary surface level emotion understanding by using visual/acoustic/distance information without any customized knowledge and visualizing the final emotion in shape-color-size diagram. In the experiment employing special demonstration scenarios where a company employee's emotion is understood by a secretary eye robot equipped with Kinect, it is confirmed that the proposed method provides deep level emotion understanding different from ordinary surface level emotion understanding. The proposal is being planned to be applied to a part of the emotion understanding module in the demonstration experiments of an ongoing robotics research project entitled "Multi-Agent Fuzzy Atmosfield".

View full abstract

Human behaviors have large influences on their own health and emotion. They have both aspects, good and bad one. To know these aspects made our daily life much more fruitful. Many of previous researches estimate and extract behavioral information using several sensors. But it is hard to extract some behavioral and emotional information using only sensory data. The authors focused on Twitter which spread rapidly these years. The users post their own behavioral and emotional information directly in the text format which is called tweet. We can extract users' behavioral and emotional information in real-time because the user posts at the moment when they are doing and feeling them. The handy characteristics of Twitter that is easy to post using a personal computer or a smartphone showed us the possibility of extracting human informal and real intention. In this study, we tried to extract behavioral and emotional information using not only sensory data but also Twitter. In addition, we analyze the influence on emotion by behavior using extracted information. We can apply this approach individually to know their personal tendencies of their thinking, feeling and doing.

View full abstract

Recently, wordplays are considered as a study dealing with interestingness of language on a computer. Some kind of rhetoric are producible using concept database. This paper briefly look at the conceptional structure of prosopopoeia, hyperbole and so on.

View full abstract

We have many kinds of data of time series, which are understood via linguistic expressions by humans. We have proposed a method to express their global trend and local features in a natural language. In the previous research, however, linguistic expression does not explain some of time series, because all time series use fuzzy sets of the same definition for three fuzzy intervals in the temporal axis. When humans define some fuzzy intervals, they usually take the big turning point into account. In this paper, we propose a method to divide time series into some dusters and define fuzzy intervals based on the turning point by grouping similar tendency of adjacent dusters. We show that the global trends using this method explain time series more appropriately.

View full abstract

We have many kinds of data of time series such as stock prices. We understand time series with a natural language. We have proposed methods to express a global trend, local features and oscillation of time series in a natural language. When the data are similar to each other in the standard deviation and frequency of oscillation, our method can't express their oscillation well for some data. If the frequency of oscillation of data is low, we seem to understand oscillation with its state, the frequency, amplitude and phase of oscillation. If it is high, we seem to understand oscillation with the upper outline of the plotting, i.e., plotting of parts of local maximums of the data, and the lower outline of the plotting, i.e., plotting of parts of local minimums of the data. We propose a method for high frequency of oscillation to express the outline of the plotting with global trends and local features of local maximums and local minimums. We express the high frequency oscillation data with this method and verify its effectiveness.

View full abstract

Our daily lives are a series of decision-making, it is an important element of life. Decision is made on the basis of the evaluation of the subject. Therefore, in our daily, we are living surrounded by the evaluation. Further, it can be considered to be living in the continuous rating. In the field evaluation problems, fuzzy measure, fuzzy integrals have been used so far. However, realistically, it is about what has been applied fuzzy integral and fuzzy measure is applied how many problems in the field, it seems to be no much yet. The reason for that one, and there has not yet accepted the concept fuzzy measure, the fuzzy integral is understood in the general public. In this paper, from the perspective as described above, it is turning to practical application to the site fuzzy measure and fuzzy integral, as a methodology for problem solving social, for the application fuzzy measure and fuzzy integral consider. Then, it is intended to indicate a new starting point in this area. In this paper, as one idea that inspired Simon to "bounded rationality", we propose a concept of "evaluation of bounded rationality".

View full abstract

Fuzzy measures and fuzzy integrals have high affinity with the Kansei engineering. But there are not so many applications on sites, because fuzzy integrals are behind on theoretical aspects. In this presentation, we discuss how to apply the fuzzy integrals to many problems on sites in the Kansei evaluations.

View full abstract

The Fuzzy-Set Concurrent Rating Method (FCR method) was introduced to measure subjects' attitudes more naturally than the traditional rating methods. Using set function representations, this paper presents the fundamental idea to obtain integrated values and expression degrees of one's opinion by FCR scores. The set functions consist of 'positive,' 'negative', 'both positive and negative', and the rest. The integrated values of the FCR-method are the Shapley values of the set function. In addition, we present the method to analyze the pair questions and show the Kansei indicators.

View full abstract

To survey the consumer psychology on purchasing high-class brand products, this paper targets on undergraduate students, and presents a new data analysis, in which the data is obtained from FCR-method and the set function representation is also used. First, we obtain data from undergraduate students by referring to the questionnaire in [2]. Second, five main factors are derived by analyzing data from FCR-method. It is shown that the first obtained factor would be quality of the brand, but the remaining ones are related with purchasing psychology. Third, using Set Function Representation in [5], Questions 17 and 4 that are contradiction for one expression are analyzed and integrated values are obtained. It is shown that ambivalence of purchasing high-class brand products is reasonable and could be explained theoretically and practically.

View full abstract

In this paper, we try to examine to what extent the company creed affects the profit. The company creed certainly contains an important message, because it is related to the root of the company. If it is possible to obtain the core factors in the company creeds, we can see what creed is better for the company. To distinguish between company creeds, we used the information-theoretic self-organizing map called "gradual information maximization". The method is good at extracting a few important variables with better visualization. By using this method, we could extract one important word, namely, "society". The companies with higher profit tend to have the company creed containing the word "society". This result suggests that the "social" factor plays a more important role in the company creed.

View full abstract

This study aims to develop a quality evaluation system by analyzing the skills of experienced puffer cooks from well-established intermediary wholesalers in order to create a model for quality evaluation. A total of 80 natural tiger puffers dressed by the cooks (Migaki) were graded into five classes to determine what aspects of the surface color and meat freshness of the fish the cooks use for evaluation. Analysis of the appearance evaluation by cooks, combined with information on fish coloration and meat freshness, indicated that the cooks focus on four colors on the surface of the fish.

View full abstract

In this paper, the models which enable the veriufication of the species by the color of the puffer meat is proposed. A total of 185 fish dressed by the cooks (Migaki) were examined to determine what aspects of the surface color of the fish the cooks use for veriufication. Fuzzy inference models were then constructed and evaluated using the colors as antecedent-part variables, and differentiation accuracy according to the models became over 90%.

View full abstract

The authors analyzed the relationship between the early stage color of the leaves of ginkgo and the progress of turning yellow. In addition, we were also compared with the decrease in the temperature environment around the Ginkgo at that time. We did an observation in "Tama New Town" of the suburban in Tokyo. The samples of the leaves were extracted from the Ginkgo biloba which is planted into the "Tama New Town" street and the Tama Art University. We were used Hue of the HSV color system for representation of color of the leaf. And we made a mathematical model using the Hue of leaves to express a progression process of turning yellow. As a result, we have shown that the difference between the initial hue of the leaves of the ginkgo is related to the speed of change of the color of the leaf in autumn.

View full abstract

A fuzzy set representation method of Kansei Texture and its visualization method are proposed to express individual difference of Kansei Texture feelings. If these are applied to online shopping, it is expected that they provide buyers the criteria if sample sending request is necessary or not according to the variance degree of individual difference and it also offers sellers the possibility information of return-goods in the case of big individual difference. A subjective evaluation experiment by 17 subjects and 20 goods shows that the correlation coefficient of the degree of individual difference and the visualization is 0.89(P<0.05), i.e., directly proportional relationship is observed between two degrees, and the validity of the visualization is confirmed. The proposals are planning to be implemented in real world online shopping and medical treatment education.

View full abstract

It is necessary for human beings to create "neo-fuzzy" concept in order to facilitate unimpeded communication with machines. The new concept is defined in the non-metric space and acts as a breakthrough for the traditional "fuzzy" concept. This panel discussion may excavate the following subjects; (1) inference, association, analogy and meaning understanding, (2) text understanding, text production, mechanism of communication with linguistic terms, (3) building process of neural networks in the brain which produce/understand "neo-fuzzy" linguistic terms, (4) corresponding mechanism of visual/acoustic Vorstellung to "neo-fuzzy" linguistic terms, (5) memory devices, system architecture, and some other hardware systems peculiar to "neo-fuzzy" information processing, (6) applications of communication with "neo-fuzzy" linguistic terms.

View full abstract