◎Akihiko Akao，Yuya Kobayashi，Sho Shirasaka（The University of Tokyo），Kiyoshi Kotani（PRESTO, Japan Science and Technology Agency），Yasuhiko Jimbo（The University of Tokyo）

Collective rhythm which emerges from neuronal population are widely observed in brain activities. Multi-regional interactions such as thalamocortical loop causes conduction delay between neuronal populations and the functionalities of such delay on brain rhythm have been experimentally suggested. However, mathematical analysis of such system was restricted due to its infinite dimensionality. Here, we analyze the dynamics of delay-coupled population of modified theta neuron. We apply a dimension-reduction technique called Ott-Antonsen ansatz for neurons with quenched variability and derive mean field dynamics as a set of delay differential equations. We analyze the emergence of collective rhythm and synchronization properties (e.g. phase response function) in the mean field model.

◎Masatoshi Hashimoto，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

A ``basin of attraction'' is a set of solutions arriving at the same local optimal solution by Best-improvement Local Search. By utilizing the concept of basin of attraction in the solution space, the solution space is interpreted as a higher structure, which is a set of basin of attraction, and a lower structure, which is a set of solutions, in this paper. Based on the hierarchical interpretation of the solution space and basic strategy in metaheuristics, an optimization method with search strategy to find a basin of attraction to which superior local optimal solution belongs is proposed. The search performance of this method was evaluated through numerical experiments. In addition, the search situation and influence on search by parameter of this method are considered.

◎Yuki Miyake，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

Recently, the development of high-performance metaheuristics has become an important subject. In this study, an adaptive Cuckoo Search based on ranking of search point is proposed. This study aims to improve the performance of Cuckoo Search by adjusting the parameter β to allow search points with good evaluation value to search nearby and those with poor evaluation value to search far away. Finally, the performance of the proposed method is evaluated by numerical experiments.

◎Wang Hongran，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

We analyzed the search characteristics of Firefly Algorithm (FA), which has a fundamental nature of a Superior Solution Set Search Problem, previously defined in our previous study for single-objective optimization problems. In this study, we proposed a new FA method based on the former problem. This method, which employs cluster information by K-means clustering, is tested for performance by fundamental numerical experiments.

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

Maintaining high level voltage stability is difficult in the power systems with contingencies. Moreover, the maintaining becomes rather difficult by high renewable energy penetration. Therefore, Cooperative Co-Evolutionary Differential Evolution (CCDE) is used in this paper. CCDE is one of the meta-heuristics and needs the decomposed power system. The convergence performance and search range of CCDE are affected by the size of decomposition, then the coexisting of the rather high convergence performance and the rather global search range is very difficult. In the proposed method, the coexisting is achieved with merging each decomposition.The effectiveness of the proposed method for emergency voltage and reactive power control is verified by simulation.

◎Hiroki Kimura，Atsushi Okuyama（Tokai University）

A multi-agent system (MAS) comprises several autonomous agents and expresses global functions based on local interactions. In recent years, approaches examining network structures based on graph theory have gained attention in MAS control theory. This study deals with the average consensus problem wherein the consensus value is the average of the initial values of each agent, using small robots with omni-wheel as agents. Notably, it studies the average consensus problem under realistic conditions and limitations with respect to communication range and number of communication units. Using simulation, the study determines the number of communication units required for achieving average consensus when applying graph theory-based control theory.

◎Wen Song，Weo Weng，Shigeru Fujimura（Waseda University）

This research is mainly about the data analysis in factories of process industries. The main target is to mine from the whole data, try to de-noise the signal and detect the potential abnormal value from the signals. Since the data is filled with noise and delays among different kind of the signals, we first use the cross-correlation and wavelet transformation to remove the delay and noise. Then, use deep-learning method to train the model with processed data, and finally use the model to detect potential abnormal value.

◎Yiyong He，Wei Weng，Shigeru Fujimura（Waseda University）

Flexible job-shop scheduling problem (FJSP) is an extension of the classical job-shop scheduling problem. FJSP allows an operation to be processed by several different machines. FJSP with overlapping in operations means that an operation is divided into multiple sublots and each sublot is treated separately and transferred to the next processing stage once it is completed. In previous research, a genetic algorithm is proposed to solve this problem. In this study, we try to improve the procedure of previous research to achieve better results. The proposed improvements were tested on a set of benchmark problems and compared with the results obtained by previous research.

◎Zhenqi Liu，Wei Weng，Shigeru Fujimura（Waseda University）

In manufacture system, process planning and scheduling are two procedures of great significance to the efficiency. In this paper, we are going to solve problems integrating process planning with scheduling (IPPS) with multi-objectives. The objectives including makespan, total machine work load, and maximal machine work load. To solve this kind of problem, the authors proposed some improvements based on the existed solutions, which is turned to be efficient in practice.

◎Tasuku Nakajima，Nao Onji，Takanori Kurao，Masaaki Fukuhara（Tokai University）

A Hamming distance search CAROM with symmetric NOR architecture has been considered. Since the Hamming distance search CAROM can compare quickly stored data and input data in parallel, and find stored data with Hamming distance within certain range, the CAROM is effective in the fields including AI (Artificial Intelligence), IoT (Internet of Things) and Big Data, which require high-speed searching. The CAROM can perform interrogation operations stably without the use of dummy capacitors. Since details of characteristics about the CAROM is not clarify, we analyze and consider with focus on power consumption in this paper. We have confirmed power consumption of the CAROM using HSPICE simulation. HSPICE simulations are performed detector designed by Hibikino 2µm CMOS process design rule.

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

In this paper, an improved teaching–learning-based optimization algorithm (TLBO) is proposed to solve the job shop scheduling problem(JSP). JSP is a strongly NP-hard combinatorial optimization problem. It is difficult to solve the problem to the optimum in a reasonable time. TLBO has a considerable potential when compared to the best-known heuristic algorithms for scheduling problems. To make TLBO become more suitable for solving JSP, a new coding method is used. New learners are introduced and special local search operators are incorporated into TLBO to balance the diversification and intensification. To show effectiveness of the improved TLBO, numerical results based on some benchmark instances and the comparisons with the traditional TLBO and the best known lower bounds are provided.

Jiarong Li，◎Qianji Hao，Wei Weng，Shigeru Fujimura（Waseda University）

A particle swarm optimization (PSO) method using new pbest list and gbest list to select pbest and gbest for Multi-Objective Flexible Job Shop Scheduling Problems is proposed in this paper. To get the better performance and better diversity and convergence, some strategies to support this new method are also proposed. The performance of the proposed method is evaluated by comparing with Classic GA, PSO+SA and PSO+TS. The computational results have shown that the proposed method is efficient and effective for solving the multi-objective Flexible Flow Shop Scheduling Problems.

◎Xuanqi Liu，Hidenori Sasaki，Hajime Igarashi（Hokkaido University）

In this presentation, the topology optimization of the rotor of a synchronous reluctance motor using the particle swarm optimization (PSO) will be discussed. The rotor shape is represented by means of linear combination of the normalized Gaussian functions. The weighting coefficients associated with the Gaussian functions are determined so that the fitness becomes maximum under the given constraint. The convergence speed, resultant optimal solution obtained by PSO is compared with those of the genetic algorithm.

◎Shingo Hiruma，Hajime Igarashi（Hokkaido University）

We introduce a new algorithm for deriving the continued fraction from a given transfer function using self-adjoint Lanczos algorithm. A Cauer equivalent circuit is directly synthesized from the system equation by this method because the continued fraction corresponds to the Cauer circuit. In the conventional method based on the Padé via Lanczos algorithm, the transfer function is approximately represented by the rational function. Therefore, to obtain the Cauer circuit, it is necessary to compute the continued fraction expansion of the rational function. Using the proposed method, the Cauer circuit can be synthesized directly from the system equation. In the presentation, we discuss application of the proposed method to the finite element equation of electromagnetic fields.

◎Yasuhito Oshima，Takuya Kinoshita，Kazushige Koiwai，Toru Yamamoto（Hiroshima University）

Physically handicapped people have been increased, on the other hand the number of caregiver has been decreased because of the declining birthrate. Therefore, this problem causes the insufficient aid for the handicapped people. To solve this problem, the some of assistive products are developed for a self-reliance life of physically handicapped people. However, those products need to tune parameters by a few of professional caregivers because of tuning difficulty for the purchaser. Therefore, the development of the assistive products adjusted adaptively is needed. In this study, a data-driven ankle foot orthosis system is developed. The system configuration of the hardware and the controller is introduced. Moreover, the control result of the system is verified numerically.

◎Daisuke Yamatani，Mitsuyuki Saito，Daijiro Yoshimura，Wataru Wakita，Yasuhide Kobayashi（Hiroshima City University）

State feedback control is premised on the ability to observe all state. But, all state cannot necessarily be observed by failure of a sensor. Since the output feedback control proposed until contains a nonlinear equation in solvability conditions as solution in such a case, a feedback gain serves as a semi-optimal solution. We have inquired about the conditions in which this output feedback control has an optimal solution. We are as a result the solvability conditions containing the Riccati equation which carried out the same form as the Riccati equation which are the conditions of the optimality of an optimal regulator. In this paper, the circle criterion of the form similar to the circle criterion known as robustness of an optimal regulator can be derived from this Riccati equation.

◎Yu Yoshida，Masakazu Suzuki（Tokai University）

Robots are required to plan autonomous actions like a human and to execute them. For executing complex actions, it is necessary to adequately determine many control parameters. In addition, the robots have to select adequate actions from many choices for unexperienced tasks. We describe the effectiveness of optimal action selection in autonomous tool operation task executions based on the ICAC and the easiness criterion. As a demonstrative example the hockey shoot task is considered. A humanoid robot executes a tool operation task by changing the way to hit a ball depending on positions of the ball and the goal.

◎Masanori Honma，Takeyoshi Nakano，Khoa Vu Van，Shigeru Takayama（Ritsumeikan University）

Telemetric Sensing Node Network System(TSNNS) is one of availability to monitor Landslide Disaster (LD). This network consists of Local Sensing Node Network System(LSNNS), Host System(HS) and Cloud System. LSNNS is the network system of sensing nodes distributed on dangerous slope of mountain or hill. LSNNS monitors a sudden movement of the slope autonomously. HS manages to monitor occurrence of LD at a remote place. By the data from LSNNS, HS estimates the movement, shows the condition and alarms the occurrence of LD to observer. On the other hand, HS also manages TSNNS to care the condition and characteristic functions. HS and LSNNS communicate data and command mutually in dual way by Cloud System. This shows the measurement methods, system operation and telemetric monitoring interface.

◎Junki Amano，Takayama Shigeru（Ritsumeikan University）

Health care is an indispensable element of Quality of Life (QOL) for ourselves and families. Telemetric Body Area Network (TBAN) System is one of the useful methods to realize the remote monitoring of daily behaviors and health condition of user. TBAN is constructed by two main system; Local Wearable Sensing Node Network System(LWSNNS), Host System (HS). They communicate data and command by E-mail and Storage service of Cloud System. HS has a role to monitor the user remotely to observer such as his family and home doctor by using Active User Interface (AUI). AUI operates various kinds of function for showing not only physical parameters but intuitive and useful information by the fusion of data collected from LWSNNS. This paper describes design and operation of Active User Interface.

◎Takeshi Isomoto，Kenji Suyama（Tokyo Denki University）

In this paper, a design method for an optimal directivity pattern of a microphone array is studied.Under an echoic environment, it is desirable to design the directivity pattern so as to minimize the maximum sidelobe level.This design can be achieved by only a fixed beamforming technique.Then, an accelerated redesign of the directivity pattern is required to track the target sound source moving.In the proposed method, the directivity pattern is adjusted based on the pattern before the move of the target sound source.Several design examples are shown to present the effectiveness of the proposed method.

◎Kenzo Ymamoto，Kenji Suyama（Tokyo Denki University）

In this paper, a design method of IIR (Infinite Impulse Response) filters using PSO (Particle Swarm Optimization) is studied. It is required to enumerate a lot of local minima for the IIR filter design problem.In this study, the particles are shifted toward another local minimum solutions by changing the desired frequency response when the stagnation occurs. As a result, the local minimum stagnation can be avoided and a lot of local minimum solutions are enumerated.Several design examples are shown to present the effectiveness of the proposed method.

◎Takashi Higasa，Shinichi Ito，Momoyo Ito，Minoru Fukumi（Tokushima University）

This paper proposes a method to input character and/or character string in an augmented reality using gesture motion. The proposed method detects the area of character string using gesture motion. The proposed method consists of three phases; hand area detection, tips of first and second finger detection and gesture detection to specify the area character. The hand area is detected by extracting skin color based on saturation value by using threshold processing. The tips of first and second fingers are detected by specifying the sharp angle points on edge points of hand area. The gesture is detected by calculating the distance between tips of first and second fingers because gesture is to take hold the tips of first and second finger.

◎Akihisa Hasebe，Kunihito Kato（Gifu University），Hideki Tanahashi（Gifu Prefectural Research Institute of Information Technology），Hidekazu Hirayu（Gifu Prefect. Gov., Dep. of Commerce, Industry and Labor）

This paper describes an automatic detection method for small dents on a metal plate. We adopted the photometric stereo as a three-dimensional measurement method, which has advantages in terms of low cost and short measurement time. In addition, a high precision measurement system was realized by using an 18bit camera. Furthermore, the small dent on the surface of the metal plate is detected by the inner product of the measured normal vectors using photometric stereo. Finally, the effectiveness of our method was confirmed by detection experiments.

◎Tomoki Yamada（Gifu University），Mariko Hosoe（Gifu Prefectural Police），Kunihito Kato，Kazuhiko Yamamoto（Gifu University）

Handwriting analysis is a method to determine an unknown writer by comparing known writer’s characters with unknown writer’s characters. The lack of characters of the same class is one of problems with the accuracy of handwriting analysis. We tried to generate different class of characters possessing handwriting features to solve the lack of characters. In this paper, we tried the learning method using deep learning to model the handwriting feature extraction and the handwriting character data generative model. We tried to learn the generative model of the data of Japanese character (Hiragana). In this paper, we propose a modified method to improve character generation accuracy by adding a convolution model to existing neural network model.

◎Kazuki Hiramoto，Kazuhiko Hamamoto（Tokai University）

This research conducts a comparative experiment using a cognitive behavior called “Reaching” to investigate a difference of human cognitive behavior (in other word, ecological cognition) in VR using HMD and the real environment and to examine a method to compensate the difference. First, a subject estimates the longest distance to a cylinder placed in front of the subject at which the subject can reach the cylinder. After that, the subject grabs the cylinder and places it as far as possible. The result is investigated in the real and the virtual environment. In the real, it has been found that the longest distance to an object for a person to decide to be able to reach for and grab the object is about 1.1 times as long as the longest distance where a person can actually reach for grab the object. On the other hand, this research shows the value in the virtual environment is not the same as one in the real. Especially, in the simplified VR environment, a standard deviation of the value among subjects is larger. And it is also shown that the standard deviation in the simplified VR environment without the virtual environment experience tends to be more larger. The results suggest the amount of environmental information and experience of the virtual environment may influence the ecological cognition in VR.

◎Hirohisa Omae，Kazuhiko Hamamoto（Tokai University），Shinji Shimosako（Pharma Mirai），Takaki Higashi（Cyber Communications），Shigeta Ikuji（Access）

In resent, the Ministry of Health, Labour and Welfare of Japan has discussion of “Health Support Pharmacy” and “administration cot of primary care pharmacist” for new pharmacy operations with a patient information unified management. In these discussion, aftertime, it is to be expected to become important to establish a patient information utilization. In this study, we product pharmacy support tool using prescription data and national land numerical Information from the challenges of a previous research.

Shigeo Wada，◎Hiroto Nishiki（Tokyo Denki University）

CBIR and OBIR are powerful tools to search image. Desired images can be obtained from image database without text indexing. We have previously proposed an OBIR using k-means clustering algorithm to reduce influence of background complexity and additive various image noise. In this paper, to improve robustness of the OBIR system, we propose an object image retrieval system using SLIC and multiple image features. Our method can retrieve the same object with a query object image in noisy and complicated background without learning process. To show the effectiveness of our method, experimental retrieval by cross-validation was executed using image database containing both background and object. Precision rate was used to evaluate the performance of image retrieval.

◎Abdelrahman Harras，Akinori Tsuji，Kenji Terada（Tokushima University）

Automatic detection and counting moving objects are essential technique in the field of computer vision used for many applications such as counting large number of objects in buildings, roads, and railway stations, etc. The purpose of this study is to detect and track moving passersby by using motion-based tracking. The background subtraction algorithm based on a Gaussian model is applied to detect the passersby. In addition, morphological operations are applied to eliminate noise and the Kalman filter is used to determine a motion of each tracked passerby. Our developed system used a USB camera fixed at the entrance of hall in a laboratory environment to obtain the image sequence. From the experimental results, our system achieved the detection ratio is 94% and the tracking is 88 % correct in case of one passerby. The detection ratio is 69% and the tracking is 61% correct in case of two passersby.

Mengbo You，◎Haitian Sun，Takuya Akashi（Iwate University）

Frontal face detection has been developed mature enough to be applied to everyday life while it is still difficult to handle various profile face views practically. A common solution is to build different detectors organized by a decision tree and ensure each detector coping with a view. However, collecting training images for each view costs much time and labor. Most profile face detectors cannot perform as good as the frontal face detector for insufficient training data. We proposed to transform a frontal face detector to a multi-view face detector without further data collection or training. The result verified that our method is able to detect human faces in almost all the head poses in the dataset.

◎Rebeka Sultana，Masahiro Noumi，Gosuke Ohashi（Shizuoka University）

Space Tethered Autonomous Robotic Satellite-Cube (STARS-C) transmits jpeg image from space in packet format which is required to be reconstructed as soon as possible. But due to high speed communication or instrumental error, all the packets are not received by the receiver. Moreover, a packet contains extra parts along with the image data that must be discarded. Also, jpeg image cannot be reconstructed with incomplete common segments. To solve these problems, we propose an image reconstruction method where we have used an ideal image to complement those common segments and padded other missing data with zeroes. The image has been captured by the same camera. The purpose of this study is, to reconstruct satellite images quickly and efficiently from packets even if some packets are missing.

◎Anqing Jiang，Osamu Yoshie（Waseda University）

Online communication become popular with the development of Internet. It includes debate, social network communication, E-learning. During the online communication, participant will show their opinion, and participants usually makes group with each other to strengthen their point of view. In this research, we the groups which contain opinion-similar participants are called opinion groups. In previous research, nobody thinking the emotion influence of each participant. If a participant agrees to others opinion, he will usually have positive emotion. Emotion is an essential feature that detection each participants’ opinions. According the emotion and topic we can know which speaker have the same opinion with others.

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

In this paper we would like to propose a method for designing a speech recognition system, targeted for non-native speakers of some particular language.The method is based on detecting and recognizing the most common speech features, amongst certain nationality speaking the different language, than its mother tongue.Having performed the extraction of most typical features from the non-native speech database, the system uses multidimensional clustering, factor analysis and sensitivity analysis to create clusters related to age, gender, speaking rate etc., and containing models used in particular cases for more robust non-native speech recognition.In this research, we used the database containing recordings of Japanese people reading sentences in English.