◎Ryuta Oosumi，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

In this paper, we formulate a new definition for single-objective optimization problem and propose Firefly Algorithm-based optimization method for the formulated problem. The formulated problem requires a set of local optima with evaluation values better than a certain level and the distances between each local optimum are farther than a certain distance (the superior solution set). In other words, the superior solution set meets the aspiration level of the user and has diversity.In addition, the Firefly Algorithm-based optimization method was developed based on an analysis of the Firefly Algorithm search mechanism. Finally, the performance of the developed method was verified by comparison of developed method with Firefly Algorithm and Particle Swarm Optimization.

◎Seijun Morita，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

Multi-objective optimization is a very active area of research. In this study, we propose a new multi-objective optimization method based on the functional specialization based search strategy. The functional specialization based search strategy is based on two ideas. The first idea is the state evaluation and classification of search points to realize an advanced search structure. The second idea is to use operations with different features to achieve an efficient search by taking advantage of each operation. The proposed method takes advantage of this search strategy to achieve high convergence and a well-diversified solution set during multi-objective optimization. The performance of the proposed method was verified by a numerical simulation using typical benchmark problems.

◎Toshiki Shimizu，Takashi Okamoto，Seiichi Koakutsu（Chiba University）

Radial basis function (RBF) networks is used for function approximation, classification and so on. Generally ,RBF networks having many basis functions has superb performance compared to not having many basis functions.However, RBF networks having many basis functions has problem that it take high calculation costs on learning and operation. For example, it will become a problem that offer service that construct RBF networks for each user and process large quantity of data by using those. Therefore, it is important that changing networks scale more appropriate with keeping performance. In this paper, we propose a method in order to consolidate basis functions without significant degradation in performance. Results of computational experiments indicate the validity of the proposed method.

◎Naoki Kawamoto，Satoshi Takayama，Atsushi Ishigame（Osaka Prefecture University）

There is a possibility that breakers of transmission lines would open because of accidents that occurred to power systems by contact of lightning strikes or wildlife. In this case, voltage collapse could occur because of shortage of active power in power systems. For this reason, we have investigated a voltage control called Emergency Voltage and Reactive Power Control (Emergency VQC) for maintaining voltage stability. Moreover, meta-heuristics is used to calculate amount of reactive power compensation equipment that enable voltage stability to maintain in the power system after failure. Specifically, we choose Particle Swarm Optimization (PSO) among meta-heuristics. We propose improvement of the PSO by Electrical Distance. The effectiveness of the method is verified by simulation.

◎Kenzo Yamamoto，kenji Suyama（Tokyo Denki University）

In this paper, a design method of IIR filters using PSO is studied. In our previous work, an avoidance method of local minimum stagnation has been proposed, in which a penalty is added to an objective function. In this study, a novel method which repeats a diversification and an intensification alternatively is proposed for a single-swarm PSO, in which a penalty range is fixed to a small value. Several design examples are shown to present the effectiveness of the proposed method.

◎Haruna Aimi，Kenji Suyama（Tokyo Denki University）

In this study, IIR (Infinite Impulse Response) filters having a null frequency in the stop band region are designed using PSO (Particle Swarm Optimization). A new penalty function for forming a null is introduced to the objective function in addition to the conventional penalty function to ensure the stability of IIR filters. Such a constraint brings increments of the local minima in the solution space. Therefore, it is important to avoid the local minimum stagnation. We design IIR filters using PSO that reallocates particles when a local minimum stagnation occurs and verify the effectiveness of the method through several design examples.

◎Linna Li，Wei Weng，Shigeru Fujimura（Waseda University）

TLBO algorithm is used to solve job shop problems. In the original TLBO algorithm, there are only the teacher and learner phases. In the teacher phase, all the students learn from one teacher and in the learner phase carry out among the existing students. So it’s not easy to find the optimal solution. To increase the diversity, firstly we introduce the multi-learning method in the teacher phase; secondly we improve the learning method of the learning phase; thirdly, improve the decoding procedure; fourthly, we combine the TLBO algorithm with local search technology. To show the efficiency of the improved TLBO, the simulation results for benchmark problems are compared with results derived by the other algorithms.

◎Yifan Ye，Shigeru Fujimura，Wei Weng（Waseda University）

The optimization of supply chain allocation problem is to design an optimal supply chain system which minimize the total system cost, and guarantee the customer satisfactory rate in certain level. In the static supply chain design problem, there are certain number of factories, distribution centers and customer zones, in which all kinds of costs and customers’ demand are certain. While in the dynamic environment, all parameters are changing in every period, and we use the mean and variance to evaluate performance of the system. As there are more than two objectives in this problem, we use NSGA-II to evaluate and compare the fitness of each solution. In order to make full use of NSGA-II, we introduced combination priority-based decoding to the problem to calculate the decision parameters.

◎Qian Shen，Shigeru Fujimura，Wei Weng（Waseda University）

The resource constrained project scheduling problem is an NP-hard problem, and an effective genetic algorithm is proposed for solving this problem. In this problem, the precedence constrained between each activity and resource constraint for each resource should be considered at the same time. The objective of this research is to minimize the total project time with these two constraints.

◎Ryota Furuoka，Shimpei Matsumoto（Hiroshima Institute of Technology）

To develop a flexible scheduling solution, rule based algorithm using on-site worker's empirical operational knowledge would be effective. Especially, IF-THEN rule based solution method is considered to be the most manageable knowledge model for on-site worker. On the other hand, currently the effectiveness of IF-THEN knowledge has been evaluated by only on-site worker's subjective view, therefore, a quantitative evaluation method of IF-THEN knowledge has been required. We have been addressed to develop a method for acquiring IF-THEN rule based scheduling knowledge and also for numerically evaluating these effectiveness. In this paper, we show a framework to evaluate the scalability of scheduling knowledge expressed as IF-THEN rule with single player Monte-Carlo tree search.

◎Toshihide Shigechi，Moe Sugimoto，Takuya Matsumoto，Hisashi Tamaki（Kobe University）

In this research, racing kart driving is taken as an example of the human-machine systems, and a simulation model, with respect to a driving agent, is designed. In designing a driving agent, information processing is considered as consisting of four stages of observation, recognition, decision and operation. In this research, ways of realizing the manner of information processing by human driving on a certain granularity and also, the framework of the decision process is newly designed and implemented by reconsidering human driving. Through some simulation results using the modified model, the validity and the potential for understanding drivers' decision-making and estimating thier skill of the model is investigated.

◎Mari Ishii，Takuya Matsumoto，Hisashi Tamaki（Kobe University）

As the performance of computers improve, it becomes possible to simulate larger-scale and more detailed traffic flows. In this research, a road traffic system has been modeled as a hierarchical autonomous decentralized system by constructing three kinds of model correspond to roads, vehicles and drivers. By modeling vehicles and drivers separately, it is possible to consider independently physical dynamics and issues related to information and decision-makings. This paper focuses on the driving decision-making model. To improve the accuracy (or reality) of the decision-making and the operation of a human driver, the part of the model concerning the steering operation is reconsidered. Moreover, it is confirmed that the revised model can reveal more human-like behavior than the previous one.

◎Hirohisa Oomae（Tokai University）

In the recent years, technology used by AR has attracted attention in various fields, such as entertainment, direction induction and medical treatment. There are expected as applications of work support in the real environment. But, since general AR contents is projected on 2D screen in spite of 3D information, to grasp sense of distance is difficult. The purpose of this study, to develop a new method to view CG part stereoscopically by using naked eye 3D tablet.

◎Kohei Oyama（Tokai University）

It is commonly believed that current door mirrors cause problem of “forward carelessness” associated with the eye movement. However, technology to project information of side- view mirrors in the forward field of view is not yet. To solve the forward carelessness problem, here I study what happens when actually there is information of side-view mirrors in the forward field of view by using the Immersive Virtual Environment to create a “virtual side-view mirrors”. Here I found that eye movement by the virtual side-view mirrors becomes easy. In addition, it indicates that virtual side-view mirrors is beneficial than current side-view mirrors in term of safety. These results provide to the future development of the mirror-less car.

◎Nobuto Fujimoto，Shimpei Matsumoto，Masaru Teranishi，Takeno Hidetoshi（Hiroshima Institute of Technology）

In recent Japan, many elementary and junior high school students feel awkward consciousness strongly to manufacturing class, and this is considered as one of the causes of the aversion to science and engineering. One of this cause is considered to be the shortage of teacher’s manufacturing experience. The purpose of this study is to develop a virtual reality blended brush coating skill training system to provide manufacturing experience such as the implicit skill which cannot be gained only by the conventional teaching materials. This paper prepares four instructional methods, and reports the experimental results.

◎Akihiro Makiguchi，Kazuyuki Takahashi，Kunihito Kato，Kazunori Terada（Gifu University），Takashi Hagi（Taiyo Kagaku Co.,Ltd.）

We propose a method for determining the geometric features used when people recognize categories of objects. In this study, we targeted the shrimps. At first, we extracted shape features by using Principal Component Analysis (PCA) on the feature points obtained from shrimp images. We carried it out contour and pattern, and aimed to post to subjects the sought shape features as images. In the two experiments, we determined the type of shrimp shape, and found the range acceptable to shrimp. The results of the two experiments confirmed the shape features in which the shrimps were judged by subjects.

◎Junpei Inukai，Kunihito Kato（Gifu University），David Harwood，Larry S. Davis（University of Maryland）

Generalized Hough Transform (GHT) can robustly detect generic objects against noise and occlusions. However, GHT has a weak point in detecting object shape change. Thus, we introduced a training process to determine the voting weight of GHT by using PLS regression analysis. We call our method “PLS Hough transform”. To confirm the effectiveness of our method, we applied PLS Hough transform to the vehicle detection from aerial and satellite images. However, there are many errors at the portion where the edge points are concentrated, such as the satellite images in the forests. As a solution to this problem, we introduce edge determination to PLS Hough Transform. As a result, this method has improved the accuracy than the previous method.

◎Hiroto Nishiki，Shigeo Wada（Tokyo Denki University）

In this paper, we propose a noise robust similar object image retrievalmethod. Robust local mesh pattern and Canny edge detector features areextracted from object image to generate histogram features. To show theeffectiveness of our method, experimental various noisy image retrieval bycross-validation was executed using three standard object image database(Coil-100, NEMA-CT image database, Essex Face DB). Recall,precision rates and F-measure were used to evaluate the performance of imageretrieval.

◎Ming Li，Takashi Komuro（Saitama University）

In this paper, we propose a method for accurately estimating the distance of a preceding vehicle using high-speed stereo video images. Using image information with high temporal density, we expect that more accurate estimation becomes possible. The relative distance of the preceding vehicle is estimated using a particle filter. By taking a weighted average of estimated states of many particles, accurate distance estimation which exceeds the limit due to image resolution is possible. We measured the RMSEs of the relative distance of a preceding vehicle by using artificially-generated video images with different frame rates. The results showed that as the frame rate becomes higher, the estimation accuracy also becomes higher when the distance of the preceding vehicle is far.

◎Takaaki Inoue，Keiko Sato（Kagawa University）

Red-green dichromats are difficult to distinguish between reddish and greenish colors. However, it is known that color naming behaviors of red-green dichromats are similar to that of normal trichromats. It has been reported that the color naming response for red-green dichromats under the condition in which color stimuli are presented simultaneously was more stable than the serial presentation in which single color is presented one after the other. In this paper, for red-green dichromats and normal trichromats, we conducted color naming experiments under two presentation conditions and four brightness conditions. Based on the collected naming data, we explored whether each conditions affected a stability of naming responses between red-green dichromats and normal trichromats.

◎Kenta Omiya，Kenji Suyama（Tokyo Denki University）

In this paper, a method for multiple sound source tracking by two microphones is studied. In our previous work, the summed MUSIC (MUltiple SIgnal Classification) spectrum was used as an evaluation function for PSO (Particle Swarm Optimization). Then, the penalty function is added for avoiding, the same direction tracking. In the proposed method, the penalty function is calculated from the mixed Cauchy distribution fitted to the denominator of MUSIC spectrum.The effectiveness of the proposed method is shown by several experimental results.

◎Masashi Matsumoto，Tomonori Izumi（Tokyo Denki University）

Tipping-bucket raingauge is generally used for rainfall observation. However it cannot observe short time rainfall by its observation principle. The purpose of this study is observation of short time rainfall by analyzing the rain sound. In the prior study, the rainfall that was observed by the sound occurred an error by effect of the wind. Therefore they corrected it using value of tipping-bucket raingauge. We aim high precision rainfall observation without correcting. We proposed the new recording method. First, we recorded the sound and observed the rainfall at the same time, in order to investigate relation between rain sound and rainfall. Then, we examine the possibility of rainfall observation at our method from this relation.

◎Yuki Nakagawa，Shigeo Wada（Tokyo Denki University）

In this paper, we propose a PCA-LDA based diagnosis method for lung disease based on breath interval extraction. We have extended our previous approach to multiple lung disease case. Three types of lung diseases are assumed.We evaluated diagnosis performance compared with conventional approaches such as MFCC-GMM and PCA-SVM based methods. Our method showed efficacy with respect to diagnosis accuracy and computational efficiency in simulations.

◎Yannan Chang，Osamu Yoshie（Waseda University）

As an essential way of human-computer cognitions, increasing attention has been directed to the study of Speech Emotion Recognition (SER). Although advances have been made in research of SER, a challenging problem is that how to choose efficient features for this task is difficult. In our research, based on a deep learning method of convolutional neural networks (CNN), we propose to involve both of pooling feature-maps and convolutional feature-maps into the fully connected layer to learn Discriminative Emotional Features (DEF). In our research, rough features are learned in the lower layers, emotion-specific features are extracted in the higher layers, and support vector machine (SVM) is used to make the classification.

◎Aya Shirai，Tota Mizuno，Naoaki Itakura，Kazuyuki Mito（The University of Electro-Communications）

Innervation Zone (IZ) location shifts according to change in muscle length, and it’s impossible to evaluate muscle contraction around IZ. This study proposed a new evaluation method for muscle function while avoiding IZ automatically, and evaluated muscular fatigue during dynamic contraction using electrode array. In the method, the characteristic waves were extracted, and IZ location was estimated using cross-correlation and amplitude ratio between a neighboring channel. The muscular fatigue was evaluated from muscle fiber conduction velocity (MFCV) using channels without IZ location. As the result, MFCV value decreased with muscular fatigue and the characteristic agree with conventional. Thus, it is possible to evaluate muscle function with changing muscle length using the system.

◎Kanako Mikami，Kunihiko Oura（Kokushikan University）

This paper discusses the change of brain blood flow and inner temperature in prefrontal cortex during attentional control task. Hemodynamics in the form of oxidized hemoglobin concentration (Oxy-Hb) is measured by Near-infrared Spectroscopy (NIRS). Some reports evaluate activation of the auditory attentional function by fMRI during Clinical Assessment for Attention (CAT), but few reports use NIRS. In our previous researches, we reported that Oxy-Hb increased in left DLPFC area during Paced Auditory Serial Addition Test (PASAT). Generally, heat is produced by the metabolism typically in brain active area when oxygen and glucose are used by brain activity. We will also consider the possibility to measure temperature change in brain due to nerve activity by Magnetic Resonance Imaging (MRI).

◎Junichiro Kagita，Yasue Mitsukura（Keio University）

In this paper, we aim to detect migraine severity by analyzing the difference in electroencephalogram (EEG) within a migraineur. Migraine is a chronic neurological disorder characterized by episodes of headache. In present medical care, severity of migraine is evaluated using the Visual Analogue Scale having the problem that only subjective evaluation is available. Therefore, an objective evaluation is necessary to make further accurate therapeutics. To solve the problem we propose an approach using EEG which is able to reflect objective bio-marker. For the experiment, a migraineur's EEG was measured undergoing 3 different states using a simple EEG device. We will be able to establish the possibility of EEG measured by a simple EEG device having the ability to detect migraine severity.

◎Shintaro Imai，Suguru Kanoga，Yasue Mitsukura（Keio University）

In this paper, the electroencephalogram (EEG) was used to evaluate comfortableness brought about by seat angle. Seat angle has a large effect on driver’s fatigue, which is one of the main causes of accidents. Finding comfortable angles while driving leads to preventing the accidents. However, there is no quantitative method to measure such angles; therefore, an alternative method of evaluation is necessary. Hence, EEG under comfortable angles and normal state were recorded at fourteen locations on the scalp, and power spectrum and cross-bispectrum analysis were carried out. Largest difference could be seen in the cross-bispectrum value of Fp1 and Fp2, suggesting the effect of comfortable angles appeared the most in the area.

◎Shuhei Takayama，Yuko Nakakuki，Takuya Matsumoto，Hisashi Tamaki，Harumi Takahasi（Kobe University），Yasuhiro Sunaga，Hideo Yokota（Riken）

Recently, the number of diabetic patients is increasing. While it is considered that impairment of insulin secretion causes such type-2 diabetes, the mechanism of insulin secretion in pancreatic beta cells is still unclear. In our research, in order to have a deeper knowledge on the mechanism, simulation-based approach has been adopted. An improved simulation model of insulin granule dynamics is proposed based on bio-imaging data and medical background. Here, a keynote point is to represent the outer layer by introducing a module structure. Through some simulation experiments by using the proposed model, it is confirmed that the results are consistent with the medical knowledge currently obtained. This shows the validity and the potential of the proposed model.

◎Rusli Muhammad Azary，Hussin Sumayyah，Shigeru Takayama（Ritsumeikan University）

Improving Quality of Life (QOL) has long been an absolute goal nowadays. The understanding, measurement and evaluation of human’s vital signs and their applications are essential in order to prevent any complications, which may reduce one’s QOL. This study presents a concrete application of measurement health-related QOL with an autonomous Wireless Body Area Network System (WBANs). A method of evaluating vital signs is performed and linked to physical intensity assistance in exercise. Specifically, BAN acts as a supportive system which can assist a user in monitoring user’s vital parameters, providing real-time feedbacks and dynamically sharing information from any location to one or more users. This paper presents the measurement of exercise intensity in jog training and stabilization.

◎Kenji Hayashi，Nur Atiqah Binti Azlan，Shigeru Takayama（Ritsumeikan University）

Wearable Body Area Network (BAN) system is one of the useful methods to realize improving user’s body behavior and monitoring health condition. Not only showing physical parameters, by feedback of vital information to user, BAN system is extended to the autonomous system closed to user. Getting such information, user is able to improve his body behavior consciously. Moreover, telemetric function realizes to inform user’s physical condition dynamically to his family or home doctor far from him. To install these functions to BAN, it is necessary to design flexible and robust data flows among Sensor nodes, Hub system, Local Human Interface and Remote Host System. This paper describes the design of dynamical data flows and management of the BAN system for real time physical monitoring.

◎Takeyoshi Nakano，Khoa Vu Van，Priyanka Supekar，Shigeru Takayama（Ritsumeikan University）

This paper shows study on telemetric sensing network system for landslide disaster surveillance. The sensing network system consists of Local Sensing Node Network System (LSNNS) and Host System (HS). LSNNS is placed at monitoring area (slope of mountains and hills). And HS is placed at a remote location. LSNNS is a network of distributed sensing nodes. The host system manages the operation of LSNNS remotely and monitors landslide disasters. The characteristics of telemetric sensing network system are autonomous operation function of LSNNS and remote management function by HS using dual way communication. This paper introduces functions of sensing nodes for observing, sensing and communicating autonomously. Finally, this paper shows the effectiveness of autonomous functions.

Xiaoyu Li，○Osamu Yoshie（Waseda University）

A great challenge in Internet of Things (IoTs) is how to protect private information from disclosing. There are many advanced methods to keep an IoTs system connect safely with the external network. However, these efforts are ineffective to avoid privacy disclosure if a malicious wireless device is put in the perceptual layer of IoTs. Therefore, we derive an innovative method called Horizontal Hierarchy Slicing (HHS) method for detecting the actual number of nodes in the IoTs topology. The privacy protection of the perceptual layer could be implemented by determining whether there are malicious wireless devices. Received Signal Strength Indicator (RSSI) data are the network parameters which are utilized in this method. The simulation results show their effectiveness in privacy protection.

◎ShiQing Yan，Osamu Yoshie（Waseda University）

Nowadays, on-line debate is getting more and more influential. In fact, most of online debates are not concerning participants’ benefit, they join the debate with social purpose. In such kind of debates, participants try to know others’ feeling and find an opinion group for himself/herself. In this paper, We propose to build bi-directional belief graph to show the belief status of all the participants, as well as a novel method to divide participants into opposite groups and neutral group if exist

◎Koki Okimoto，Shimpei Matsumoto，Shuichi Yamagishi（Hiroshima Institute of Technology），Tomoko Kashima（Kindai University）

The objective of this paper is to examine the causes of preventing novice programming learner's understanding from the programming education of reading source codes. The authors developed a learning support system for source code reading. The developed system was utilized in a programming lecture for novice programmer. This paper obtained student responses from a questionnaire and the learning log data after the students had completed one semester of the instruction in programming, and analyzed these data. The analysis result clarified the description which may make program reading comprehension difficult for programming beginners.

◎Kacper Radzikowski，Le Wang，Osamu Yoshie（Waseda University）

In this paper we proposed a method for correction of non-native English speakers' speech, based on the speech's domain focused document.Our approach consists of three important steps. Detailed document analysis in the preprocessing phase, finding the document part corresponding to recognized by SR sentence, and sentence correction part.The analysis creates the BM25F method lemmas' vectors, and the keyword vectors.Finding the corresponding part is realised using the HMM based method.The correction is conducted by calculating the final probability connecting keywords comparison, BM25F lemmas comparison and HMM based method, for each candidate sentence, and choosing the most probable.The candidate sentences are obtained from the result of speech recognition software.

◎Ritsuko Daimon，Akinori Hagioita，Tomoaki Tanaka，Takashi Tomii（Yokohama National University）

Recently, many digital archives are published to preserve and share precious cultural properties. Nevertheless, these archives are often constructed for experts, and ordinary people are difficult to understand the value of these properties. The Kosode Byobu collection is such a precious cultural property, and shows clothing and fashion culture of the Edo period. The collection is too profound to understand for ordinal people. Therefore, it is necessary to construct a browsing system in order to help ordinary people understand the value of the collection. Accordingly, we constructed a system for browsing the Kosode Byobu collection. As a result, it has become able to see changes of the fashion in the Edo Period and explanations of the fabric, at a glance.

◎Qiuhao Fu，Kazushige Koiwai，Toru Yamamoto（Hiroshima University）

As a common and effective approach, PID control has been widely used in many systems. However, in some cases, it is difficult to use the PID controller to control a system, which has a long time-delay or a high-order lag element. Therefore, controllers with more complex structures are required for these systems. In this paper, a new controller design method which can decide both the control structure and the control parameters at the same time is proposed. The control structure is decided by the genetic algorithm, and the control parameters are calculated by using the least squares method. In this way, a controller that can obtain a desired control performance is designed. Furthermore, some numerical examples will be verified.

◎Xinwei Liu，Hiroki Sato，Shigetaka Takagi（Tokyo Institute of Technology）

High slew rate, low power operational amplifiers (Op-amps) are bottlenecks of today’s battery-powered equipment, due to the limited tail current and stability. To overcome this problem, this paper presents a novel slew rate boost scheme. It can gather extra tail current by slew rate boost circuit and decrease the value of compensation capacitor by using capacitance multiplier compensation when output voltage is slewing. Cadence Spectre simulation shows that this op-amp has a slew rate of 105 V/µs, consuming 0.68mW.