Biomedical Fuzzy System Association (BMFSA) was established in May of 1985 and the first annual meeting was held in February of 1989. Many presidents of BMFSA, many chairmen and program chiefs of annual meetings, chief editors of journals and many earnest members contributed to progress of BMFSA and development of the soft computing sciences. Unfortunately it is said that BMFSA is coming to "the corner" at the age of twenty years old. From the view point of looking back the history and forecasting the future, I tried to survey the activities of many contributors to twenty years history of BMFSA, through the journals and the paper books of national annual meetings. The analyzed result are summarized into four categories : A. Method of the paper, B. Purpose of the paper, C. Field of the paper and D. Stage of the work.

View full abstract

After presenting a basic outline of the Semantic Web, it is described possible roles for fuzzy logic, with its use in approximate search and matching, and as part of the knowledge representation, including a fuzzy ontology structure.

View full abstract

[in Japanese]

View full abstract

We have proposed "functional type SIRMs method" in [2] in which the consequent part of ordinary SIRMs method is generalized to a function. In this paper, learning algorithm of the proposed SIRMs method is derived by steepest descent method, and is shown to be superior to ordinary fuzzy reasoning methods by applying to a diabetes diagnostic system.

View full abstract

We examined that effect of repeated listening to wind band music for heart rate in healthy human at habitation or place-of-work environment. By making wind band music (only one piece for five times) was heard, four healthy human's heart rate was measured. It turned out that heart rate's value is decreased or hardly changed as the number of listening to wind band music was increased. On the other hand, it was shown that change tendency of heart rate in two men was similar. And same result was also obtained in two women.

View full abstract

This paper describes a concept of decision making support system using DBA. The system using Web and XML technology is aimed to make the decision dynamically and effective. It is hopeful to make human decision making from the Information selected by method using DBA.

View full abstract

The authors recorded nineteen-channel event-related potentials (ERPs) during recognition of two types of Japanese character ; Kanji (Chinese characters) and Hiragana (one type of phonetic characters). By field-sequential stereoscopic 3D display with liquid crystal shutter, a word and a non-word were simultaneously and independently presented to the left (right) eye and right (left) one, respectively. Each word consists of two Kanji or three Hiragana characters. Three subjects were instructed to press a button when understanding the meaning of the visual stimuli after 3000 ms poststimulus. Equivalent current dipole source localization (ECDL) with three unconstrained ECD was applied to the ERPs. For both Kanji and Hiragana, the ECDs were localized at the occipital and inferior temporal gyrus at 200 400ms, and those at the Wernicke area at 300 600ms. The latter ECD for one left-handed subject was localized at the homologue of Wernicke area.

View full abstract

In this paper, we suggest knowledge acquisition considering coverage index, and appraise precision of rules by fuzzy discriminant analysis. Aditionally we compare precision of ordinary method to our method.

View full abstract

A Genetic Algorithm (GA) is a search algorithm based on the mechanics of natural selection and natural genetics. Searching in GA is achieved by iterating reproduction, crossover and mutation strategies. In these strategies, the reproduction strategy is the most important for deciding the searching points of the next generation. However, traditional reproduction operators may lose genetic diversity of population in an early stage, because they can not generate new chromosomes which are different from the present chromosomes. A decrease of genetic diversity leads to ineffectiveness of searching. We have proposed the reproduction strategy by utilizing an approximation ability of a binary version of a Self-Organizing Map to maintain the genetic diversity of the population. However, the searching performance is influenced by an ordering of inverting elements (bits) of a binary weight vector. In this paper, we propose a modified reproduction strategy with new updating rule of binarv weieht vectors based on schema.

View full abstract

TAM (Topographic Attentive Mapping) network is a biologically-motivated model with Gabor function type receptive fields. In this paper, we discuss a new TAM network structure with extensive receptive fields and a restructuring method of fuzzy knowledge acquired from the network structure. We also show the usefulness of TAM nework using some examples.

View full abstract

An Index of arteriosclerosis adopt Stiffriessparameter-β (β-value) that is measured by a blood vessel picture of carotid artery. A diagnosis is made by carotid artery. However, this diagnosis is not easy task.. Then, We construct diagnosis support system that β-value is forecasted the dangerous factors of arteriosclerosis and β-value of arm artery (easy diagnosis).

View full abstract

A library provides us with books, journals, news papers and so on according to the demand of its users. The selection of books to buy is also done based on the mission of each library. Usually, they purchase books according to their object as a city, an institute, a school, a university, etc. are founded. It is more important to install books which users are requiring. In this paper we provide a method how to adjust selection of books to the demand of users. The objective of the paper is to propose a method for strategic decision of selecting books. A method is also to deal with latitude in an aspiration level of the decision maker according to their mission. In this method, a librarian and a manager define, for each of expected utility rate and its variance, a necessity level and a sufficient level of users. And then, they can obtain a solution that satisfies an aspiration level of the users. We employ this method to analyze and decide which kind of books and how much rate of the total budget should be spent to buy, where the method of fuzzy mean-variance analysis is employed to solve the problem.

View full abstract

Recently, Hospital support systems are rapidly being computerized using such as Diagnosis Reservation System, Ordering System for Drug Prescriptions, Electronic Medical Record (EMR) and so on. These data, however, is not used effectively for analysis because of the great deal of the data and its variety - image data, numerical data, text data and so on. Previously, knowledge discovery methods from numerical medical database were studied and discussed actively. Text mining method for medical data, however, has not been enough discussed. This study discusses knowledge discovery method from Electronic Medical Records using Self Organizing Map (SOM) and its possibility. This paper proposes the brief sentence mapping method using SOM from Incident Reports. This paper describes the keyword extraction method from Incident Reports and the coding method of SOM.

View full abstract

In order to develop a data mining system for huge database mainly composed of numerical attributes, there exists necessary process to decide valid quantization of the numerical attributes. Though the clustering algorithm can provide useful information for the quantization problem, it is difficult to formulate appropriate clusters for rule extraction in terms of cluster size and shape. In this paper, we propose a new method of quantitative association rule extraction that can quantize the attribute by applying clustering algorithm and extract rules simultaneously. From the results of numerical experiments using benchmark data, the method is found to be promised for actual applications.

View full abstract

In this article, we describe about an analysis of fundamental visibility on Web color sets by Self-Organizing Maps (SOM), and an paired comparison test for its training. In recent years, people in many age groups use the Web frequently. However, there are few Web contents in consideration of their accessibilities. Therefore, there are also many pages that are hard to see for people in a specific age group (elderly people). We think that the visibility of the Web will be very important in the near future. As visibility, we focused only on the combination of character and background colors, which is called color sets, among the design-elements of Web contents. In the combination of character and background colors, we conducted the paired comparison test for collecting the standard data of visibility, and analyzed the paired comparison data using the Self-Organization Maps.

View full abstract

[in Japanese]

View full abstract

Reinforcement learning is a promising approach to realize intelligent agent such as autonomous mobile robots. The paper describes hierarchical reinforcement learning that memory based learning algorithm is combined with Profit-Sharing reinforcement learning algorithm hierarchically in multi-agent environment. By using hierarchical reinforcement learning, this research aims at learning the behavior of the agent in the hunter problem, i.e. reinforcement learning problem in the multi-agent environment.

View full abstract

The modern medicine is very effective to mending the sickness. On the other hand, the ancient chinese medicine is effective to the health promotion so as not to become to sickness. Therefore, modern medicine and ancient Chinese medicine are necessary for us to send a healthy, happy life. In this paper, we propose a home health care system based on the modern medicine and ancient Chinese medicine. This system will be called a home doctor system. Because most symptoms like the pain or tiredness, etc. are vague concepts, these vague concepts are shown here by fuzzy set in the home doctor system. Moreover, the proposed distance type fuzzy reasoning method is used as a reasoning method.

View full abstract

A novel membership function and a fuzzy clustering derived from a viewpoint of iteratively reweighted least square (IRLS) techniques resolves the problem of singularity in the regular fuzzy c-means (FCM) clustering. A FCM classifier using the membership function and Mahalanobis distances makes class memberships of outliers less clear-cut, which thus resolve the problem of classification based on normal populations or normal mixtures. Computational experiments show high classification performance and compression rate on several well-known benchmark data sets.

View full abstract

In this paper, an orthognathic surgery support system, in which an esthetic score of postoperative profile is considered, is proposed. In ordinary operation planning, an esthetic score is estimated from predictive postoperative profile. However, it is difficult to predict the postoperative profile with accuracy. In the proposed system, classification maps of preoperative profile and hard tissue movement are achieved using multiple discriminant analysis, and reasonable hard tissue movement can be estimated by observing these two maps.

View full abstract

The machine learning is a method to be effective in the non-linear multivariate statistical analysis. These methods are affected by the sample size, because the machine learning suppose there are many samples in the data. As an example of machine learning, we use a feed-forward neural network, and show that it is more effectively than a conventional statistical analysis such as Fisher's discriminant analysis and quadratic discriminant analysis. Through the simulated data, we also investigate that the machine learning algorithm is affected by the sample size.

View full abstract

[in Japanese]

View full abstract

In general, as to the fuzzy node fuzzy graph drawing, it is not easy to find an optimal solution. However, by applying the fuzzy Shapley value and the genetic algorithm, we could effectively draw the fuzzy node fozzy graph on a latticed plane.

View full abstract

We show the importance of Information Science on the progress of the life science by taking an example of biological structural aspects of the Type Strain by the 16S rDNA. First we discuss the number of DNA which are picked up by Primers. Although known primers are usually around 20 bases, we cut the line of the bases and count the number of the DNA including the primer among 4386DNA. Secondly we search so called "Universal Primer". The Universal means that the array commonly exists among three domains of bacteria, Archaea and Eucarya, however, it is not so easy task to find the universal common domain. Showing several results, it is concluded that the main reasons we could not find the complete Primer is due to the noise among the data.

View full abstract

Though various contents are provided through the internet recently, it is not easy to collect favorite contents among huge amounts of contents in terms of use's preference. In this paper, we focus on the collaborative filtering algorithm in the recommender system. We propose a fuzzy modeling approach for collaborative filtering. In our approach, valid simplified fuzzy model is constructed through optimization of MAE (Mean Absolute Error) and ROC (Receiver Operating Characteristic).The model decides the weights of similarity from the value of correlation, age difference of users, and etc. Through numerical experiments using MovieLens data, the approach is found to be promised for improvement of filtering model accuracy.

View full abstract

We measured electroencephalogram (EEGs) in recalling long term memories (Method of Loci) consolidated by mnemonics. We analyzed brain activities using the equivalent current dipole localization (ECDL). ECDs were localized to the left supramarginal gyrus (SMG), to the middle temporal gyrus (MTG) called Wernicke's area, and to the left middle frontal gyrus (MFG) called Broca's area. After the Broca's area, ECDs were localized to the prefrontal area (PFA), and then to the MFG and the inferior frontal gyrus (IFG) of both hemispheres. In comparison with the previous fMRI studies by Ichihashi et al., some ECDs were localized to the same areas. Furthermore, we clarified the transition of brain activities by the time series in recalling task by the Method of Loci from the presented sentence.

View full abstract

In the present article, we propose a feed-forward neural network model for data of mutually exclusive groups and a set of predictor variables. We illustrate some applications of the bootstrapping which implement the bootstrap-based information criterion when selecting the optimum number of hidden units for a neural network model. We examine the thyroid disease database that is not only a pure training sample study which compares estimated measures of predictive performance, but test sample which reports the realized test sample error rates associated with a constructed prediction rule. In addition, influential analysis is also investigated based on deviance.

View full abstract

This paper discusses fuzzy reasoning method by Shu in which the consequence B' with difference for a fuzzy rule A⇒B and a given fuzzy input A' is inferred according to the spatial relationship of fuzzy sets A, A' and B. Especially, learning method of multifold fuzzy reasoning is investigated by justifying scaling factors of difference with which inference results are obtained from fuzzy rules, and shows satisfactory learning results.

View full abstract

There exist many inverse problems in various fields, e.g. CT in medical field, inverse dynamics in control field, and so on. Puzzle problem is also one of our familiar inverse problems. It is important technique to solve such problems as optimization problems. Though GA is one of promising approaches in order to solve inverse problem through optimization, we should solve "early convergence" problem and "diversity maintaining" problem. This paper presents a DGA applying different search method to each group (island) with selective migration strategy. We discuss the performance especially resulting from the selective manner of migration, through numerical experiments of illustration logic problem, in terms of accuracy and diversity of the solutions.

View full abstract

This article discusses feature extraction by Symbolized particles store type neuron model including reaction mechanism which has a simplified analogy of the function that real neurons and synapses do by chemical particles. The new type of neuron model is proposed including its simulated synaptic function where the symbolized particles are generated, stored and transmitted toward the next layer. The simulated system of the model has a reaction mechanism in simulated synaptic region would be shown some strategy. The feature extraction of the information technology engineers examination is performed using this model.

View full abstract

The usefulness of clinical regional brain blood-flow measurement has sufficient evidence. There is a report on the examination for the measurement technique in PET and SPECT. Although the self-sustaining artery collecting-blood method of MS model is accurate, the invasiveness is strong for a patient, and inspection implementation technique is complicating and it keeps the staff busy with artery stop bleeding. Although many methods of vein blood collecting rectifying have been examined, nobody has not expanded to clinical application Then, we have stopped setting time series fuzzy hypothesis model in vivo model, and examined if we could calculate the regional brain blood flow by date mining the measurable plural factors observation and time series. As a view, we regard data containing an arterial blood related factor as perfect data, and the data with a loss of this factor as imperfect data. Then, we conducted the data-mining simulation by the data from inspection. We considered the imperfect data with many factors and chose the cluster-thinking as algorithm attributed to the perfect date. In other words, we adopted the algorithm as not the method SOM but the method BP. The result with having artificial intelligence has shown that the imperfect data is attributable enough to the perfect data. With the result and enough numbers of patients, we carried out the multi-variable analysis. We have found very high correlation between rCBF value and rCBF value which used the prediction function of a multi-variable-analysis result had very high correlation. It was suggested that the partial brain blood-flow value computed by this venous blood rectifying method could sufficiently be used for clinical.Reproducibility in this case and the necessity for verification to the difference between institutions remain.

View full abstract

In this article, we describe a Web browsing method with an automatic color correction (palette changing) for personal user's visibility. Especially we describe a paired comparison test and SOM (Self-Organizing Map) analysis for achromatic character and chromatic background colors, and we propose a color set prediction method with SOM for personal visibility on Web browsing. We are implementing and evaluating a web application system for this method as CGI (Common Gateway Interface) software on a HTTP server of the Internet. Since a user profiling is required beforehand for every user, we have to consider about this problem with SOM. The proposed method is available for this problem by the 3 conditions of the prediction method, which are "both-ends colors", "forward colors", and "backward colors".

View full abstract

Recently, it became very attractive to measure the health degree by the pulse wave measuring method. Then, we can analyze the degree of arteriosclerosis, and the vein age by the subject's pulse without visiting medical doctors. Already, we proposed the system using the Self-Organizing Map (SOM) for finding the health and the fatigue degree where the measuring time is very short. In this report, we propose the technology for raising the analysis accuracy from the statistics distribution obtained from the pulse wave.

View full abstract

Face recognition is an important technology in computer vision because of its applicability to various computer systems such as man-machine interface and security system. In order to develop the face recognition system, estimation technique of face direction can contribute to get robust performance of recognition. In this paper, we discuss about development of face recognition system based on estimation of face direction that can achieve independent performance of background scene and color of the face in the image. From the results of numerical experiments, the estimation technique if found to be promised for development of actual face recognition system.

View full abstract

For instance when the outputs are expected to be linear, non-linear outputs are obtained from fuzzy reasoning using the MEN-MAX method. Those errors become more serious especially in the case of multi-input. This presentation will illustrate the problem using computer analysis and analyze the most critical area mathematically.

View full abstract