Osteoporosis is a condition of fragile bone caused by the decrease of bone quantity and the degradation of bone trabecula. In the diagnosis of osteoporosis, a quantitative evaluation of bone strength is important. Bone quantity is evaluated quantitatively by using an X-ray or ultrasound method. However, bone trabecula is visually evaluated for the moment only by observing an X-ray image. A method to evaluate bone trabecula quantitatively is therefore desired. In this study, we propose a method for aiding the diagnosis of osteoporosis by using fractal analysis of phalangeal CR image. In the proposed method, regions of interest (ROIs) are automatically extracted and the bone trabecular patterns of ROIs are quantitatively evaluated by the fractal analysis. In the experiments, the differences of fractal dimension between the normal and the osteoporosis patients were confirmed.

View full abstract

In recent years, the application of augmented reality technologies heve been studied in various fields. One of these applicable fields of AR is the medical field. In our laboratory, we have studied AR system which supports an operation by laying the image of ultrasonic diagnostic equipment on top of the real world. However, the conventional recognition technique using AR marker has some problems. For example, this method takes a long time for installing AR markers and does not correspond to occlusion. Therefore, in this study, we propose the system and its algorithm that recognize space only with a camera without using AR markers.

View full abstract

Creating a virtual reality (VR) in computers by combining visual information and pseudo touch sensing through the haptic point-and-response device is actively investigated. However, it is not easy for visually impaired individuals to sense a virtual reality relying solely on audio-haptic sensing, being strictly detached from visual information. The purpose of our project is the development of haptic VR technologies for aiding visually impaired individuals recognize his or her surroundings and sense geometric constructions. We plan to investigate creation of VR sensing by a visually impaired individual from the pseudo haptic sensing at the fingers through the data-gloves worn on them. We show that it can be applied to increase physical securities of visually impaired individuals, for example, on the pavement without tactile blocks where he could be got lost or on the railway station where he is faced with a danger of falling down onto the railway. Our method is to inform him through the hand-held long cane a virtual haptic sense of guiding tactile blocks or of a guide-fence on the platform against falling-down.

View full abstract

Currently, the number of Total Knee Arthroplasty (TKA) in Japan is 70,000. Because it is easy to expect increasing the number due to super-aging society, the clinical outcome of TKA surgery has been important. The surgery requires proper soft tissue balance to avoid shortening fatigue life of prosthesis. Thus, a measurement system of force concentration about contacting region between femur component and polyethylene insert contributes to prolong the prosthesis. This study examined the force concentration by using our developed measurement system. This study performed two experiment consists of physical and simulated experiment. As a result, our developed system could localize the force concentration point. A future work is to apply this measurement system to the cadaveric knee.

View full abstract

We are developing a head mounted display (HMD) which aids a low vision individual in his visual recognition (we may call it an "eyesight aid" taking a visual equivalent for "hearing aid"). We assume our targeting eyesight aid to be equipped with the following helping functions.1. Visual recognition: it helps the low vision individual find out the object in his view which he wants to recognize or the one which he is better to recognize (for example, a pictogram of a toilet).2. Proper viewing: if he is of narrowed view or of lack of part of view, it navigates his view (gaze) point to the direction in which he could catch the object in the clear view region.3. Easy seeing: it enhances the image on the display, for example, it exchanges black and white colors, adjusts contrast.In this paper we describe design guidelines for developing such HMD. We also give situations where a low vision individual increases his walking and recognition of the environment by the use of the HMD.

View full abstract

Currently, papers have reported measurement systems for joint motion in diagnosing. Although measurement systems with optical device such as camera is used widely, it have disadvantage that the precision decreases in case of presence obstacle between the marker and camera. Therefore, it is not applied to a clinical scene that hands and arms of an examiner interfere the vision of the camera. This study proposes an analysis method of knee motion by extracting only the region of the femur and tibia from images taken by an optical device.

View full abstract

Tactile graphics are images that use raised surfaces so that a visually impaired person can feel them. Tactile maps are used by blind and partially sighted people when navigating around an environment; they are useful for explaining the layout of a building or outdoor environment. Tactile Map Automated Creation System (TMACS) is developed for producing tactile maps. Tactile maps, produced by TMACS, often contain many information of roads and streets. The structure of a map should be as simple as possible because of the readability of the map. TMACS, however, does not eliminate unnecessary information automatically; in a tactile map, necessary information such as the route, roads and landmarks should be presented. Thus it is difficult for the visually impaired to read a tactile map produced by TAMCS. The goal of our research is to propose a method for automatic translating hand-drawn maps to tactile maps. This is because a user can draw freely maps, so that the readability of tactile maps is improved. In this paper, we discuss a pattern recognition method for extracting and classifying objects of hand-drawn maps.

View full abstract

The anterior cruciate ligament (ACL) is one of strong ligaments in the knee. The ACL has a role to prevent excessive internal rotation and forwarding movement of the tibia. Lachman test is one of manual tests for examining the ACL. An examiner discriminates an injured degree of the ACL by feeling finger stress. Some measurement devices for Lachman test have been reported. However, these devices aim to analyze not the stress but the motion of the knee joint. Therefore, we set the goal of this study to develop a measurement system for the finger stress with force sensor during Lachman test. A quantitative measurement system during Lachman test with force sensor was developed. A basic experiment before numerical analysis of Endpoint revealed that the inclination of force can separate characteristics of string and elastic. Lachman test was measured by using this system on intra-examiner, inter-examiner and nonexpert. This result may indicate that this analysis method and the system have usefulness so that the bimodality or multi-modality were observed over half data. A future work is to apply the injured knees.

View full abstract

In this paper, a simulator of an elderly occupant is developed. The developed simulator generates the pattern of transition between multiple areas inside the house. The generation of the transition pattern is based on a typical transition pattern that is identitied from a set of observed data in a real experimental environment. The validity of the generated occupancy is investigated using some matching measures.

View full abstract

It is difficult for the visually impaired to walk outside alone. For this reason, many systems for assisting people with visual impairments when they walk outside have been studied; almost all of them are developed using GPS, RFID, and image processing. A system based on GPS does not work inside buildings, and a system based on RFID requires many infrastructures. On the other hand, image processing becomes an important technology to develop a system for assisting people with visual impairments; this is because many of the visually impaired have smartphones, and so they can take images by themselves.Base on the background above, we discuss a method for extracting character strings from scene images. In this method, the Canny edge detector is first applied to a scene image, and we then extract character strings. Many edges in a scene image are detected from non-character components. Thus, in our method, edges from non-character elements are efficiently eliminated by using dilations of mathematical morphology and local segment densities. After that character strings are extracted by fuzzy inference. Finally, computer experiments are conducted for showing the accuracy of our method.

View full abstract

Recently many smart mobile terminal have developed. The authors have studied supporting the color vision defective using such a mobile terminal. In this paper, we describe developing an application software for estimating defference in vision of the color vision defective.

View full abstract

It is important that we reduce environmental load of the whole life cycle of the product as a problem to aim at the formation of the sustainable society. In this paper, we propose the calculation method of the environmental load in consideration of the importance of each state in the life cycle.

View full abstract

This study is concerned with efficiency analysis of Profitability through data produced from a business simulation game. We concentrate on two factors, namely selecting of evaluating item and improving of range for efficiency. The efficient methods to select proper items and improve efficiency range are proposed based on the financial ratio analysis and the sensitivity analysis, respectively. The objective of this study is to propose a method that can make up for shortfalls of conventional Profitability, of financial, ratio analysis methods and the DEA model (CCR) with quantitative and non-quantitative data produced from production, sales, financing and other business operations on a business simulation game

View full abstract

In this paper, we take up a multi purpose group decision making problem and we propose a solution based on Data Envelopment Analysis (DEA) and a method of decision making hierarchy (AHP). In multi-objective problem, it is necessary to evaluate multicriteria factor for synthesizing the determination result of each standard and also it is important to weigh each of reference. Furthermore, in more than one person decision-making, it becomes a problem when each of person's opinion mix to other people's opinion. This decision overall is also related to the idea of consensus and social choice theory, but in this paper, to explore ways of determining overall subject, we use the variable weighted data envelopment analysis and pairwise comparison of hierarchical decision making method.

View full abstract

A group decision is often believed to be better than individual decision simply because of accumulating all members' knowledge, being able to check errors carefully and so on. The other reason is that the group facilitate creativity and develop new ideas by inspiring each other. For these advantages, the decision makers improve their given information and may revise their opinions referring to the others and the group's as in Delphi Method. While, this paper assumes the synergy effect of decision makers. Someone's consistency may be complemented by the others opinions and/or the integration may create the new value beyond any individual ones. Based on Interval AHP where uncertainty among comparisons are reflected into interval weights, the individual evaluations, their sum, and the group evaluations are examined.

View full abstract

Bala and Goyal (2000) showed that a center-sponsored star network is the strict Nash equilibrium in two-flow model, a mathematical model of network formation. However, the result of the laboratory experiments conducted by Falk and Kosfeld (2003) was different from analytical prediction. In the experiments, some Nash equilibrium networks were formed but any strict Nash equilibrium network was not formed. Additionally, Nash equilibrium networks were not stable. This study makes a hypothesis such that the decision making of a human subject is not always rational, but he/she makes a decision based on several information about the situation through trial and error. Based on the hypothesis, we conduct a simulation system using ZCS (Zeroth-level Classifier System), extract some behavioral rules, and analyze the behavior of the subjects.

View full abstract

In a robot production subject at universities and high schools, robot competition is employed for aiming to motivate students and to encourage intelligibility of technology and knowledge. Robot competitions have a high degree of availability for reducing problems of science course and engineering. In this presentation, we introduce the basic robot production subject at Department of Electronics and Mechatronics, Tokyo Polytechnic University. And we verify the effectiveness of robot competition in production subject using questionnaire investigation.

View full abstract

The report of Ministry of Economy, Trade and Industry shows that industry expects cultivation of embedded system developers. This study aims at propose of effective method to cultivate embedded system developers and develop it in National College of Technology. We have noted robot, one of the embedded system, as educational material. Then we propose education method using walking robots that carries I/O and controller, and can be implemented advanced information processing technology. At first, we have set higher-grade students who have already understood a procedural programming language in National College of Technology as subjects. And the exercise that needs their knowledge has been designed and conducted. We have proposed characteristic exercises because 2 types of robot, biped and quadruped, are used in this education. It is found that proposed method is useful in accomplishing objective because subjects described they can have an experience about development of embedded system in their report. In this report, the detail of conducted education method is shown. And the effect of proposed method is evaluated by subject's answers for given tasks and their feedbacks.

View full abstract

Robotics is a very large field which requires various disciplines such as mechanics, mathematics, electronics, physics and information technology. Robotics education is, therefore, becoming more and more important in engineering education at college. From the point of view of craftsmanship, an education using robots can give students more experimental results than by normal lectures. It is also expected that a robot education will be able to give them a strong impression. In this research, we propose the new lecture system called "Introduction of Robot project" and "Basic of Robot project". These classes are aiming active learning lecture. Therefore, this class uses group working, debate, and presentation, positively. And, one of the big points is that this class held for a few department students which are belonged in department of mechanical and electrical engineering, media technology and aerospace engineering, respectively. In this paper, we present the aim of this lecture system and introduce the contents of lecture.

View full abstract

In the competition of RoboCup soccer Middle Size League, development of the hardware needs various technology for making big machine. Thus, we have made various robots by improving some device every year. In this report, we introduce our soccer robot, which attend to RoboCup since 2009.

View full abstract

In the field of RoboCup Middle Size League, self-localization is important to estimate own position including goal and robot positions and to decide strategy. The self-localization method uses the internal information, such as direction detected by electronic compass, and surround information such as camera image. However, the orientation information by the electronic compass changes depending on the surrounding environment. In this paper, we verify the accuracy of the self-localization and the relationship of the accuracy of the electronic compass in RoboCup MSL soccer field.

View full abstract

In the field of RoboCup Middle Size League, self-localization is important to estimate own position including goal and robot positions and to decide strategy. The self-localization method uses the internal information, such as revolution counted by encoder and direction detected by electric compass, and surround information such as camera image. In this paper, we construct basic self-localization method, and verify it in RoboCup MSL soccer field.

View full abstract

RoboCup is an international join project to promote Artificial Intelligent (AI), robotics, and related field. It is an attempt to foster AI and Intelligent robotics research by providing standard problems where a wide variety of technologies can be integrated and examined. In RoboCup, soccer game is selected as a main topic of research, aiming at innovations to be applied for socially significant problems and industries in the future. We also have organized RoboCup team "Hibikino - Musashi" Middle-Size League soccer competition. In this paper, a new fast moving mobile platform "Musashi150", which moves maximum 3.5 m/s, and problems in the world conference is introduced.

View full abstract

Consistency Index is the most important measurement of the results from pairwise comparison in the AHP. Pairwise comparison is a method to calculate the weights for each element in order to perform a comparison of two advantages. The consistency of the answer will be determined by the value of C.I. of a pairwise comparison table. If its value is for the case exceeding 0.1 (Saaty’s criteria 1980) is not consistent, pairwise comparison is performed again. A long-standing weakness of AHP is explored byconsidering combinations of research costs, which seeks to address a difficult problem. This article aims to develop an efficient method to decrease the value of C.I. using an evolutionary algorithm, such as a genetic algorithm. Finally, we propose a software solution and discuss its effectiveness.

View full abstract

This paper analyzes two pricing schemes (linear pricing and two-part tariff) offered by a producer in which it sells a product a separate retailer who makes a cost-reducing investment before contract. Linear pricing has the advantage of providing the investment incentive for the retailer, but has the disadvantage of double marginalization. On the other hand, two-part tariff has the advantage of dissolving double marginalization, but has disadvantage of withdrawing investment incentive to the retailer. The initial marginal cost and the efficiency on investment play important roles in equilibrium. If the initial marginal cost is high and the efficiency on investment is high, the producer offers a linear pricing to its retailer.

View full abstract

In this paper, I present the simple estimation of trader’s risk averse degree expanding Roll (1984) model.

View full abstract

Sailing is a reinforcement learning problem. The only energy source of sailboat is wind. Decision making is need in partially observable and unsteady environment. State and action spaces consist of continuous variables such as appearance wind, target position and sail direction. Some of them are vectors. This paper shows reinforcement learning simulation with particle filter which aims to learn human sailor's knowledge acquisition of predicting and utilizing wind.

View full abstract

A main of this study is to develop a sign language recognition system in a first person view framework for a smooth communication tool between a hearing-impaired person and a normal hearer. As an intrinsic issue in the first person view framework, it is required to deal with frame-out of hands arising from a closeness between a camera and hands. A detection of frame-outs by skin color extraction methods, which are commonly used as a hand detection, doesn’t work well. This is because it is difficult to discriminate a hiding of hand due to another hand from a frame-out of one hand. That is why it is required to develop a method to distinguish those. In this paper, a method to distinguish the hand hiding from the hand frame-out is proposed, and then some experimental results are demonstrated.

View full abstract

Learning therapy is a kind of non-drug therapy for dementia elders. It ameliorates, stops or slows the progress of dementia symptoms. Its effectiveness is observed not only in the functional recovery of intelligence but also in that of emotion. The recovery of intelligence is quantified by indices measured by tests similar to intelligence tests. On the other hand, the recovery of emotion has been a matter of subjective impression shared by staffs in nursing homes. We, thus, attempted to quantify the recovery of emotion by measuring smile intensity. We recorded movies of elders undergoing the learning therapy across a year, and developed an image analysis program that can measure the smile intensity of elder people.

View full abstract

Refering to conventional researches such as decorating face in video image by enlaring eyes and whitening a skin to modify the face in a video to more beautiful one and its real-time implementation, this reserach aims at investigating decoration of a face depending on the subject’s attributes such as gender, age, and so on. Enumeration of attributes available for the decoration have been conducted followed by consideration of methods being able to utilize for extraction of each attribute. Methods of decoration by using the attributes have been proposed with consideration of their real-time implementation.

View full abstract

Examination of blood ow is one of the inspections concernig to lifestyle-related diseases. In this study, the red blood cell velocity is evaluated from the moving image of the conjunctival microvessel. However, it has noises such as blurring and/or blinking. In this study, the method to reduce blurring and/or blinking and then to evaluate automatically the red blood cell velocity is proposed. Applying the proposed method to an actual moving image of the conjunctival microvessel, we could successfully evaluate automatically the red blood cell velocity.

View full abstract

The major cause of acute coronary syndrome (ACS) is a blood clot generated by a rupture of unstable plaque in the coronary artery. Therefore, in a diagnosis of the ACS, it becomes very important to accurately distinguish the stable plaque and the unstable one. Several methods to classify the tissue of the coronary plaque have been proposed so far. They are not always useful however due to the performance limitations of the features selected by each method. In this study, we propose a new method to apply a sparse coding to the tissue characterization for improving its accuracy. The effectiveness of the proposed method has been verified by comparing the characterization results by the proposed method with those by the conventional methods for the real intravascular ultrasound data obtained from the human coronary arteries.

View full abstract

Automobile is changing along with a direction from an existense being merely a vehicle for moving to the one having various value-added functions. There are many examples of such direction as automatic brake by detecting forwarding obstacles, unmanned group crusing by communicating with forward and backward vehicles, and ohter safety driving and automated driving technologies. Also the use of battery of an electric car for home electricity usage with smart charging and re-charging scheduling has been investigated. As is referred by a term of "car robotics", there are many recent researches focusing on similarity of mobile robots and cars where a problem may arise that human aspect inside a cabin is not so fucused on and it is a topic to be investigated in near future. This research investigates a possibility for automovile not only a vehicle for merely moving to a destination, but also being a partner of human, by utilizing intelligent information processing techniques to watching over a car driver inside a cabin. By building a reliable relation between a car driver and a car based on communication between them, this reserach may directs methods for giving relieved to the driver, applications to enhance safety and confortability by watching over a car driver by a automated system. By enumerating functions necessary to such applications, this reserach disbusses on methods to achieve the functions.

View full abstract

The purpose of this study was identified factors to develop a sense of responsibility of project member, and investigated a subject for responsibility and a recognizable role among members using a qualitative research. This analysis showed the member's responsibility was induced by team activities.

View full abstract

For members of the project to stick to the way of the past, this paper explains the action project must take from the simulation and case. We consider the case of cell production implementation project, we propose a multi-agent simulation.

View full abstract

In general, a project organization is consisted by many kind of roles, for example, project manager, team leader, and members. Project organization charts tend use hierarchical structure. However, actual communication and decision process are complex in real projects. Informal communication pass which are not described in organization chart are very important. In this paper, we will discuss about project models of hierarchical structure organizations and Autonomous distributed organizations.

View full abstract

Tingyi (Cayman Islands) Holding Corp. (the "Company"), and its subsidiaries (the "Group") specialize in the production and distribution of instant noodles, beverages and instant food products in the People's Republic of China ("PRC"). The Group started its instant noodle business in 1992, and expanded into the instant food business and beverage business in 1996. In March 2012, the Group further expanded its beverage business, the strategic alliance between the Company and PepsiCo in the beverage business in the PRC was established. The Company is exclusively manufacturing, bottling, packaging, distributing and selling PepsiCo non-alcoholic drinks in China. The Group's three main business segments have established leading market shares in the PRC's food industry. Year-to-date December 2012, the Group was the market leader in instant noodles and ready-to-drink teas, gained 56.4% and 44.2% market shares respectively. In this paper, we try to research the success of management strategy and analyzed the VRIO analysis of the company.

View full abstract

As we all know, Chinese market is an integral part to the world, so international automobile companies are making efforts to gain access to Chinese market. To expand the market share in china, companies implement various strategies, as a result, making the market competition fiercer. As Chinese automobile companies are not well-known in the world, we consider BYD —— the representative of Chinese automobile companies. Firstly, we will summarize the history and outline of the company. Then, we will analyze the strengths and weaknesses of the company internal environment by using fuzzy VRIO (stand for value, rarity, inimitability and organization) analysis method. Applying fuzzy theory in the analysis of management environment, for instance, VRIO analysis method can make it more effective to formulate strategies. We hope that application of fuzzy theory to management methods might make a contribution to the development of Chinese automobile business in the future.

View full abstract

This paper aims at improving the usability of the book search engine for children: "Adventure in the Sea of Books", which we developed for "International Institute for Children's Literature, Osaka" in 2002. This book search engine is equipped with the content with which children can enjoy searching for books through the Internet. Each book registered in the library system is given some keywords relating to its outline or contents, along with its basic bibliographic data. These keywords enable library users to search for books by the free words with which they come up, as well as by book titles or authors. However, in order to write the outlines of books and assign keywords to them, librarians first have to read books one by one and select suitable ones for children. In that study, we confined our analysis to the bibliographic data of "picture books" installed in "Adventure in the Sea of Books". We classified the keywords assigned to each picture book into influencing ones and influenced ones, and examined the relationship between them. In this paper, we confine our analysis to the bibliographic data of "storybooks" installed in "Adventure in the Sea of Books". We examine the assigned keywords and relationship between them, and see if there is any difference between "picture books" and "storybooks" based on our previous study. Moreover, we investigate whether there is any difference in the keywords and their relationship when we classify the "storybooks" into those for the lower grades, for the third and fourth grades, and for the upper grades. We hope that this study will help to reduce librarians' workload of assigning keywords to books by hand.

View full abstract

A fuzzy autocorrelation model which we proposed is a fuzzy time-series model using a fuzzy autocorrelation coefficient. A fuzzy autocorrelation coefficient obtained by the fuzzy operation tends to increase its ambiguity. On the other hand, center values of fuzzy autocorrelation coefficients and predicted values by our model are coincided with AR model. Therefore, we can understand our model describes a possibility of a time-series system by the interval of our model. In addition this, our model is able to get describe the possibility with high accuracy. However, forecasted and predicted values of our model by using fuzzified stock prices were unnatural sometimes. In this paper, we will use fuzzy random variables to solve this problem. This is a fuzzy autocorrelation model with confidence intervals of fuzzy random data.

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 Knowledge Acquisition If the data has changed greatly.

View full abstract

Timetabling problem is a kind of combinatorial optimizations. However, it is very difficult to solve this problem with the enormous total number of combination and the complexity of the limitation condition. In this study, we develop the timetabling algorithm for Osaka International University. So, we design the reduced model and propose a solution method by Differential Evolution (DE). Specifically, we propose one of generating method of initial population. This study is a new trial such as to apply the DE to timetabling problem. Furthermore, one of new solution method is also proposed. This method creates a timetable for sophomore and freshman simultaneously. Moreover, it can treat the connection of class, such as both sophomore and freshman can attend a same class. We will report on the announcement day about the results obtained from the new proposed method.

View full abstract

We have proposed the processing methods for the fuzzy interval data which including vagueness of observer. And the usefulness of these methods are investigated. In these methods, to consider the vagueness of observed data, the correction value is added to the result which is obtained by normal statistical method using the representative values of fuzzy interval data. And as a application method, we have proposed the processing method in model selection by AIC. On the other hand, in the case which collect AIC directly, it is becoming clear that the effectiveness of correction value is hidden. And the influence for AIC by fluctuations of the number of data has been investigated. However, only investigation from a viewpoint by the mean is carried out. So, in the collection method, it is analyzed in detail that the fluctuations by the number of data in AIC.

View full abstract

This paper, we try to schedule TOC multipurpose problem with precedence relation and ambiguous resource allocation. The analyzed by DEA method completion time in the stage prior to the introduction of resources, ambiguous cases can be considered, we Precedence relation of work and allocation of resources, corresponding to the degree of satisfaction and satisfaction of each, optimum satisfaction I propose a method to calculate the completion time and degree. As a result, lead to minimization of the maximum completion time in TOC.

View full abstract

Recently, a lot of studies on Multi-Objective Genetic Algorithm (MOGA), in which Genetic Algorithm is applied to Multi-objective Optimization Problems (MOPs), have been reported actively. MOGA has been also applied to engineering design fields, then it is important not only to obtain good Pareto Solutions but also to analyze the obtained Pareto Solutions. In order to analyze Pareto solutions obtained by MOGA, it is required to consider both the objective space and the design variable space. In this paper, we define "Non-Correspondence" between the objective space and the design variable space in terms of distance relation between individuals, and extract the non-correspondence area in Pareto Solutions based on Spearman's order correlation coefficient calculated by the distance with each individual between in the objective space and in the design variable space.

View full abstract

A speciation shows some similarity to date clustering in terms that one set becomes multiple separated sets. Based on this idea, we previously proposed a data clustering method inspired by a model for speciation. The data clustering method is highly consistent with a genetic algorithm (GA) in terms of procedures, and it is expected that the method helps GA maintain diversity of population. In this paper we propose an optimization method that combines the data clustering method with GA, and evaluate the proposed method through simulations. Simulation results demonstrate that the proposed method yields better results than GA.

View full abstract

GAssist is a high-performing machine learning algorithm, which obtains a classifier through genetic learning. In this paper we examine the effects of implementing GAssist as a parallel distributed model. In our parallel distributed model, the population of individuals is divided into multiple subpopulations. Training data are also divided into multiple subsets, which are subsequently assigned one per subpopulation. In each subpopulation, a genetic algorithm is performed separately from the other subpopulations. Additionally, we rotate the training data subset used for each subpopulation periodically. In doing this, we avoid over-fitting of each subpopulation to the local training data subset. Through computational experiments we examine the effectiveness of our parallel distributed model with respect to its search ability, generalization ability and computation time.

View full abstract

Typical fuzzy reinforcement learning algorithms take value-function based approaches such as fuzzy Q-learning in MDPs and use constant or linear functions in the conclusion parts of fuzzy rules. In a previous paper, we proposed a reinforcement learning algorithm in the policy-gradient approach. Our method can deal with fuzzy sets even in the conclusion parts and also learn rule weights of fuzzy rules. This paper shows that the proposed learning method is applicable and effective to a decision making problem of a soccer robot that plays in RoboCup Soccer Small Size League by experiments.

View full abstract

Agent2d is used as a base program by many teams that participate in the RoboCup 2D Simulation League. Agent2d prepares a tree search using a series of actions called a chain action and a state evaluation function for a player agent to plan and determine the next action when the agent gets the ball. This paper realizes a return pass between a passer and a receiver in the framework of a chain action by producing better edges and nodes in the search tree and improving an evaluation function that evaluates the nodes. The new function evaluates a node, which represents a set of positions of players and the ball, from the standpoint that the passer can score if it kicks the ball towards the opponent goal at the current position. Our experimental results show the effectiveness of our proposed method.

View full abstract

In this study, we examine solutions for lemon markets using signaling by agent simulations. In lemon markets where the quality of merchandises deteriorates, the deterioration is solved by appropriate signals are provided by sellers. We notice the outbreak of quality deterioration in the market with appropriate signals. Through simulation results, we found that sellers with lower quality arise in the market and get higher profit because of their lower cost to provide merchandises.

View full abstract