| 倉角 哲也 | M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title:Development of the rehabilitation support system to quantitatively evaluate and guide training
abstract:Rehabilitation is necessary to improve physical functions, and robots are used to make training efficient. In hand rehabilitation, surface electromyogram is mainly used as a biological signal for control of robot assist hand and evaluation. However, there are few quantitative evaluations that are indexes of rehabilitation. So, we propose a quantitative evaluation method by measuring muscle shape using distance sensors. language of the presentation: Japanese | |||
| 小川 岳大 | M, 1回目発表 | 光メディアインタフェース | 向川 康博 |
|
title: Correcting temporal variation of sunlight in hyperspectral imaging of stained glass
abstract: Hyperspectral imaging contributes to analysis in various fields such as chemical substances distribution imaging. We develop a system for hyperspectral imaging by combining a high-speed rotating mirror and a spectroscope to analyze stained glass in cultural heritage. Since this system takes long time to scan the subject spatially, a temporal variation of sunlight can be a problem in the hyperspectral imaging. We explore a method to correct hyperspectral images by estimating the temporal variation of sunlight. We currently analyze the affected hyperspectral images and extract the spectral factor of sunlight in each pixel. We’ll validate if this factor is useful to correct the hyperspectral images. language of the presentation: Japanese 発表題目: 波長・空間共に高解像度なステンドグラスの計測における太陽光変動の補正 発表概要: もし、波長分解能が高い分光情報を持つ空間解像度の高い画像を取得できれば化学物質分布の可視化など様々な応用が期待できる。本研究では、文化遺産のステンドグラスを分光解析することを目的とし、高速回転ミラーシステムと分光器を組合せたシステムを用いて波長・空間共に高解像度な画像の撮影を行う。この装置では空間の走査に時間がかかるため、ステンドグラスを撮影すると、計測データが太陽光の変動の影響を受ける。そこで、変動の影響を受けたデータを解析することで太陽光スペクトルとステンドグラスの透過スペクトルに分離し、太陽光成分の変動を推定することで、計測データを補正することを検討する。 | |||
| 川上 莉穂 | M, 1回目発表 | 情報セキュリティ工学 | 林 優一 |
|
title: Study on Control of Electromagnetic Information Leakage Using Intentional Electromagnetic Wave Irradiation
TENPEST is a technique to extract confidential information from electromagnetic (EM) waves unintentionally radiated from information device. On the other hand, In the past investigations, devices with weak leakage EM waves intensity were not exposed to the threat of TEMPEST. However, in a recent study, a new threat using HT was reported, and it was shown that it is a target of TEMPEST even in devices with low EM leakage. The threat of HT causes security degradation by attackers connecting small electronic circuits to the target device. On the other hand, since such a circuit has a simple structure, there is a possibility that it exists naturally inside the device. Therefore, in this research, we use an unintentional HT structure existing in a general circuit, then we will consider a method for forcibly causing information leakage. Moreover, we will discuss countermeasures against this kind of threats. language of the presentation: Japanese 発表題目: 発表概要: | |||
| 加賀 正樹 | M, 1回目発表 | 光メディアインタフェース | 向川 康博 |
|
title: Investigating Motion Deblurring Method in Time-of-Flight Measurement
abstract: When Time-of-Flight camera measures a moving object, it cannot acquire accurate depth. This is because the influence of motion blur is convoluted with the response of light according to the depth. In this research, we propose a method to obtain an accurate depth by removing the influence of motion blur from the ToF measurement value. In this presentation, we discuss about the result of the blur removal method used in ordinary cameras can be applied to the exposure image, which is used to calculate the depth obtained by ToF measurement. language of the presentation: Japanese 発表題目: Time-of-Flight計測におけるモーションブラー除去手法の検討 発表概要: 動的な環境にTime-of-Flightカメラを用いて距離計測を行うと，誤った距離が得られてしまう．これは，モーションブラーの影響が距離に応じた光の応答に畳み込まれてしまうためである．そこで本研究では，動的な環境下でのToF計測値からモーションブラーの影響を取り除き，正確な距離を求める手法を提案する．本発表では，通常のカメラで用いられるブレ除去の手法を，ToF計測で得られた距離計算前の露光画像に対して用いられないか検討した結果を報告する． | |||
| 片山 寛基 | M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title:Development of The Tactil Proximity Sensing System Using Ultrasonic
abstract: In manipulation, the need for proximity sensing technology is increasing. In the case of using reflection of light and the case of using the change of capacitance, there are problems of environmental conditions and conditions of the target object, respectively. Therefore, in this research, we propose an ultrasonic tactile proximity sensing technology that can acquire two kinds of information, object posture and material. By applying the technology used as a sonar, we can recognize the posture of the object independently of the environment or object. Furthermore, by applying the technology used for nondestructive inspection, we obtain the material of the object. language of the presentation: Japanese | |||
| 神川 怜史 | M, 1回目発表 | 計算システムズ生物学 | 金谷 重彦 |
|
title:An analysis of eye movements during medical image interpretation
abstract:It is difficult for medical school students to learn skill to interpret medical images because the knowledge and skills required for students increased recently. To attain this difficulty, I perform to develop an educational system that makes it possible to apply an elucidated process to an application that provide feedback about learners’ mistakes in real time. It is necessary to clarify the experts’ skill of interpreting medical images. Previous studies have only studied static measurements such as gaze time and gaze frequency. This research focuses on the dynamic measurement, that is, the process of gaze. The combination of Fixation of gaze movement and Saccade will make a difference between experts and medical school students. Moreover, when making it into an application, it is possible to present the interpretation process to the learner in real time. language of the presentation:Japanese | |||
| 川合 崇大 | M, 1回目発表 | ネットワークシステム学 | 岡田 実 |
|
title : 非公開 abstract : 非公開 language of the presentation : Japanese | |||
| 倉地 亮介 | M, 1回目発表 | ソフトウェア設計学 | 飯田 元 |
|
title: Automatic generation of test code using similar code search tool
abstract: In recent years, while the requirements for software have become sophisticated and diversified, users are increasingly demanding to ensure software quality and reduce costs. In this way, the use of automatic test generation tools is progressing to ensure the quality of software and achieve cost reduction. However, the test code generated by the existing tool is far from the actual target program scenario. Therefore, developers have difficulties to understand and challenge to make maintenance work difficult. In this research, We propose a reuse method of existing test code which is practically used by using similar code. language of the presentation: Japanese | |||
| 川崎 聡大 | M, 1回目発表 | 計算システムズ生物学 | 金谷 重彦 |
|
title: Determination of Japanese Body Constitution based on daily life and Chinese Medicine (CCMG-J) Questionnaire
abstract: The importance of human health has been increasing as national healthcare expenditures have been expanding. Recently, the idea of curing “mi-byo”—pre-symptomatic diseases—has emerged, and it is reasonable to hypothesize that considering human body constitution is essential for curing mi-byo. The Constitution of Chinese Medicine Questionnaire (CCMQ), along with the Japanese version of it, are used to determine human body constitution. The questionnaire consists of 60 questions and classifies body constitutions into 9 groups. However, the questionnaire does not include questions concerning lifestyles, which could significantly affect body constitution. In addition to that, differences in body constitution between Chinese people and Japanese people have not been considered. In this study, we collected some questionnaires and analyzed the answers to the questions using hierarchical clustering. The number of questions was 332 and the number of answerers was 851. In the future, problems concerning how we group the people and how we determine their body constitution will be explored. language of the presentation: Japanese | |||
| 川端 康平 | M, 1回目発表 | ネットワークシステム学 | 岡田 実 |
|
title : A Study on Using Leaky Coaxial Cable for Wireless Power Transfer
abstract : Leaky coaxial cable (LCX) exists as a countermeasure against a difficulty for radio waves to reach places such as underground and tunnels. This cable radiates radio waves through slender holes on the surface to communicate. In our laboratory, Multiple-Input Multiple-Output (MIMO) communication and high precision position detection, using LCX have been proposed, but LCX has not been used for wireless power transfer yet. In this research, I propose to use this LCX as a power transmitting antenna for wireless power transfer. language of the presentation : Japanese | |||
| 鐘ヶ江 由佳 | M, 1回目発表 | ソフトウェア設計学 | 飯田 元 |
|
title: Investigating cases of build tool changes from OSS projects
abstract: To make software development more efficient, build tools are neccesary. Buildtools are automate tools for compiling and testing. There are many build tools with different features. Therefore, in software development projects, they select better build tools according to the attributes of the project. As the project evolves, the use environment of the build tool also changes. Therefore, in order to develop software efficiently, it is sometimes necessary to introduce a new build tool. However, the developers cannot measure an effect of the new build tool in advance and it takes time to select and change the build tool. For this reason, decisions on build tools change are difficult. In this research, in order to support judgment of build tool change, we analyze change cost of build tool and relationship between attributes of OSS projects and build tools. language of the presentation: Japanese | |||
| 木下 泰輝 | M, 1回目発表 | 知能コミュニケーション | 中村 哲 |
|
title: Measuring affective sharing between two people from simultaneously recorded EEG signals
abstract: Empathy has been studied intensively in the field of human-machine interaction and human-human interaction. The researches in this field adopt models of inter-human empathy. However, in most of these researches, empathy is not evaluated directly e.g. questionnaire-based measuring. Therefore, it is unknown whether human-machine empathy has a characteristic common to inter-human empathy. In this research, we simultaneously record Electroencephalograms (EEG) signals from two people engaged in facial expression. Then, we conduct an analysis to detect affective sharing which is one of the main components of empathy. Compared to previous research, our method doesn’t require high-cost instruments. Thus it is easy to apply our method to the field of human-machine interaction. language of the presentation: Japanese | |||
| 木俣 大樹 | M, 1回目発表 | 環境知能学 | 萩田 紀博 |
|
title: Excitement Determination Method for TV Chat Robots
abstract: In order to develop a robot that watches and chat TV programs together with users, we propose a method for determining excitement in real time from SNS comments. The time series change of SNS comment is examined, the timing when the number of comments sharply increased is defined as excitement, and the excitement level is given from the average and the distribution of past comments. By setting the appropriate excitement level, it is possible to express excited speech of a more natural robot by adjusting the speech frequency of robot. In order to verify the validity of the proposed method, we applied opportunities to football relay, we were able to judge the excitement of the opportunity and the goal scene. language of the presentation: Japanese | |||
| 久保 基 | M, 1回目発表 | 知能コミュニケーション | 中村 哲 |
|
title:Tourism Information Analysis Using Tensor Decomposition
abstract: When analyzing data, clustering is now one of the fundamental approaches. When we use the clustering, the data is mostly expressed in vector or matrix form. However, since the scale of data has been increasing, it has been diversifying and becoming higher-dimensional. Therefore there are dimensional restrictions in vectors and matrices. There is a tensor as a structure expressing high dimensional relational data. As a clustering method using tensor representations, tensor decomposition has been proposed. In this research, we propose a new knowledge discovery system regarding tourism information by applying tensor decomposition. We are currently implementing several types of a tensor decomposition, and we plan to cluster real data in the future. language of the presentation: Japanese | |||
| 二又 航介 | M, 1回目発表 | 知能コミュニケーション | 中村 哲 |
|
title: Data Augmentation thorough back-translation and factored LSTM with low resource data
abstract:Situation-aware applications must produce naturalistic outputs depending upon the environment such as modulating their speaking style or politeness. However, much resource is necessary to achieve this. In this study, we propose a new method to augment monolingual text corpus with low resource. This approach can be applied in many fields such as style transfer for dialogue system and data augmentation for machine translation. We first learn a latent representation of the input sentence using back-translation in order to better preserve the meaning. Then factored LSTM, which automatically distills the style factors in the low resource monolingual text corpus, is used to separate the intent within the context and stylistic factor. language of the presentation: Japanese | |||
| 尾﨑 諒介 | M, 1回目発表 | 自然言語処理学 | 松本 裕治 |
|
title: Automatic Stance Detection using neural network for Fake News
abstract: The popularization of SNS such as Twitter and Facebook has made it easy to obtain a large amount of information. Along with that, expansion of fake news on SNS is a social problem. In order to solve the problem, in this research we aim to automatically detect fake news using neural network and to help fact checkers. As a first step of that goal, we introduce the proposed method of stance detection using neural network. language of the presentation: Japanese | |||
| 栗田 優輝 | M, 1回目発表 | 環境知能学 | 萩田 紀博 |
|
title: Enthusiasm Analysis and Verfication for TV Chat Robot
abstract: In order to develop a robot that watches and chat TV programs together with users, we propose a method to add excitement and excitement levels from the number of posts of SNS. This makes it possible to dynamically change the amount of utterance of the robot according to the excitement section and the level, and it is considered that more natural excitement can be realized. Also, we compare the excitement of people with the excitement of SNS, and extract and characterize each feature. Moreover, by clarifying the feature promoting excitement impression, it is thought that you can properly select SNS comment which makes robot speak. As a result of subjective evaluation of the user by the robot utterance reflecting each feature, it was confirmed that the utterance of a human who contains a lot of sensuous verbs gives the impression of excitement to the user with a significant difference recognized. | |||