コロキアムB発表

日時: Wednesday, Novermber 28, Time 2 (11:00~12:30)


会場: L1

司会: 佐々木 博昭
石川 拓志 M, 1回目発表 インタラクティブメディア設計学 加藤 博一
title: 3D Pose Estimation in Sprint Scenes
abstract:Athletics sprinters require accurate 3D pose in sprint scenes to analize and improve their running posture. However, accurate 3D pose estimation from images remains a challenge. In this study, we propose model-based approach using 3D scan technology. Model-based approach focuses on estimating 3D pose using the 3D model and images of running sprinters.
language of the presentation: Japanese
 
磯貝 昇吾 M, 1回目発表 ネットワークシステム学 岡田 実
title:A receiver-side compensation circuit for mitigating no-load current in inductive power transfer systems
abstract:In the inductive power transfer (IPT) systems to which series resonance is applied to the transmitter side, there is a problem that generating large current in the transmitting coil when the receiver does not present. This will cause harm to the source, increase loss and therefore necessitate complicated turn-on/turn-off mechanism, especially in in-motion wireless charging where load appears and disappears randomly. In this research, I investigate a novel compensation topology which is implemented only at the receiver side by using three resonant capacitors. As this method is deployed only in the receiver, it keeps the transmitter current very small when the receiver does not present by exploiting large reactance of the coil; meanwhile, sufficient current will be generated automatically when the receiver is coupled to the transmitter.
language of the presentation: Japanese
 
穴井 達 M, 1回目発表 数理情報学 池田 和司☆
Title: Brain mechanisms of value acquisition, maintenance and updating through long-term conditioning and reversal conditioning
Abstract: Choosing valuable objects is critical for survival so individuals should adaptively learn better behavior depending on their located context or environment. In an unchanging environment, it may be an optimal to maintain their values for a long. If an environment changes, individuals should update their values flexibly. To make clear the neural mechanism of adaptive learning (i.e., acquisition, maintenance and updating value), we conducted the experiment that consisted of three weeks conditioning task and one-day reversal conditioning task, and analyzing both the behavioral and neuroimaging data. As a preliminary result, the conditioned stimuli assigned different values were decoded from participants’ brain activity patterns.
Language of the presentation: Japanese
 
池田 圭佑 M, 1回目発表 数理情報学 池田 和司☆
Title:The effect of functional connectivity on cognitive function
Abstract: White matter hyperintensities are considered to cause a decline in cognitive function through changes in brain networks. Recent studies suggest that changes of functional connectivity (FC) and structual connectivity (SC) contribute to decline in cognitive function. The purpose of this research is to test the hypothesis that the divergence of FC and SC causes decline in cognitive function. As a preliminary research, we verified relationship between cognitive function and some indicators obtained by applying graph theories to FC.
Language of the presentation: Japanese
 
伊藤 佑起 M, 1回目発表 数理情報学 池田 和司☆
title:Resting-state current source estimation using Graph Laplacian
abstract: MEG is one of the devices that can measure the human brain activity with the hightest temporal resolution. MEG measures the fluctuations of magnetic fields accompanying neural activity using the sensors placed around head. It is used for functional imaging and detection of epileptic abnormal brain activity. However, the problem of current source estimation using MEG generally cannot be solved because the number of estimated ones is larger than observed variables. It is required some constrains to solve the problem but methods suitable for resting-state have not been proposed yet. In my research, I propose a new method based on structure connectivity. A network of structure connectivity was applied for Graph Laplacian and analyze this eigenvalues and eigenfunctions. Resting-state networks match the spatial patterns which are Laplacian eigenfunctions corresponding to certain natural frequencies which is Laplacian eigenvalues. This eigenfunctions was named connectome harmonics. I incorporate connectome harmonics with an equation of current-source estimation. I solve the diffusion equation on the network using Graph Laplacian.
language of the presentation: Japanese
 
井上 直樹 M, 1回目発表 インタラクティブメディア設計学 加藤 博一
title: Design of Finger Motor Skill Training with Gamification based on Music Play
abstract: Some rehabilitation's task of the finger motor skill is simple. So the many patients can't go along with it. So I provide the training with gamification based on music play. The user places his hand on the device and applies the force to the finger according to the instructions indicated by the musical score. The score comes down from the top and is represented by augmented reality. The user can see it through head mounted display.
language of the presentation: Japanese
 

会場: L2

司会: 新谷 道広
谷 優里 M, 1回目発表 ユビキタスコンピューティングシステム 安本 慶一
title: Availability Investigation of Smile Sensing in QoL measurement
abstract: Daily stress has effects on mental illness and mental health, and causes a decrease in social activities. Quality of life (QOL) is an index of the quality and satisfaction with people’s daily lives. Measuring QOL can be a cue for reconsideration of daily stress and life improvement. However, QOL measurements require users to answer 100 questions defined by WHO, and thus daily measurements impose a heavy burden. On the other hand, previous psychological experiment indicated that facial expressions have several effects on emotions. In this study, we focus on the relationship between smile and QOL, investigate various methods to measure smile in our daily life to explore the availability in measuring QOL.
language of the presentation: Japanese
発表題目: QoLセンシングにおける笑顔データ活用の検討
発表概要: 日々のストレスは人間に身体的,精神的な影響を及ぼし,社会的活動を行う能力を低下させる要因となる.QOL(生活の質)は日々の生活の満足度と質を評価する指標であり,測定することで日々のストレスを見直し,生活改善のきっかけをつくることができる.しかし,QOLを測るにはWHOQOL-100と呼ばれる100個の設問に答える必要があり,日常的な測定には負担がかかる.一方で,心理学研究により表情と感情には密接な関連性があり,表情が感情に与える効果が証明されている.本研究では笑顔とQOLの関連性に注目し,日常環境での笑顔センシングについて調査すると共に,笑顔データがQoLの測定にどの程度寄与できるのかについて検討する.
 
鳥越 庸平 M, 1回目発表 ユビキタスコンピューティングシステム 安本 慶一
title: Datotsu mothion recognition using inertial sensor unit for Kendo progress support system
abstract: Recently, methods to visualize and share the performances of athletes using inertial sensor units (IMU) mounted on wearable device etc. are expanding in sports field. In this research, I focus on the traditional sports in Japan, Kendo, and propose the recognition and evaluation method of the datotsu motion during the practice swing using the IMU. In addition, I aim at improving practice efficiency and learning correct datotsu motion. As the current state of progress, SenSticks were attached to the right wrists, waist, right ankle, Shinai Tsuba, Shinai Saki-Gawa for the six subjects, and made eight type of basic swings. Based on the acceleration and gyro data obtained from this, I will classify them using a machine learning method in the future.
language of the presentation: Japanese
発表概要: 剣道上達支援システムの開発に向けた慣性センサユニットを用いた打突動作認識
発表概要: 近年,スポーツ競技において,ウェアラブル端末などに搭載されている慣性センサユニット(IMU)を用いて選手のパフォーマンスを可視化し,共有する取り組みが広がっている.本研究では,日本の伝統的なスポーツ競技である剣道に着目し,IMUを用いて,素振り中における打突動作の認識及び評価手法を提案する.これにより,練習効率の向上や正しい打突動作の習得を目指す.現在の進捗状況として,6人の被験者を対象に,右手首,ウエスト,右足首,竹刀鍔,竹刀先革にSenStickを取り付け,基本的な8種類の素振りを行った.これにより得られた加速度およびジャイロデータを基に今後機械学習手法を用いて分類を行なっていく.
 
BROU WILLIAM-FABRICE D, 中間発表 ネットワークシステム学 岡田 実, 林 優一, 東野 武史, Duong Quang Thang
title: Near-field and far-field simultaneous wireless information and power transfer systems
abstract: Simultaneous Wireless Information and Power Transfer (SWIPT) as the name suggests is a technique allowing to transmit the information and the power wirelessly. There are two (2) main techniques for SWIPT, the near-field, non-radiative using the magnetic or electric field to transmit power/information in short distance (meters) between the source coil and the receiver coil. The far-field, however, using radiative transmission, transmits power/information over long distance (kilometers) from a base station through microwaves. Both present different design challenges but more importantly share the same requirement which is the trade-off between the data rate and the energy harvested as an essential factor to evaluate the performance of a SWIPT system. In this presentation, the results of experiments on near-field proved the feasibility of the concept of the parallel line feeders as a source for SWIPT. Also, for the far-field SWIPT, a proposed joint optimization algorithm for a downlink OFDM system shows a high data rate under the constraint of high harvested energy.
language of the presentation: English