| 吉本 暁文 | 1461012: D, 中間発表 | 松本 裕治，中村 哲，新保 仁，Kevin Duh |
|
title: Coordination-aware Dependency Parsing
abstract: Coordinate structures pose difficulties in dependency parsers. We propose a set of parsing rules specifically designed to handle coordination. The new rules are compatible with existing similarity-based approaches to coordinate structure analysis. We analyzed the time complexity of the proposed rules as well as their coverage of the Penn Treebank converted to the Stanford basic dependencies. language of the presentation: Japanese | ||
| 澤井 裕一郎 | 1451054: M, 2回目発表 | 松本 裕治，中村 哲，新保 仁，進藤 裕之 |
|
title: Semantic Structure Analysis of Noun Phrases using Abstract Meaning Representation
abstract: We propose a method for semantic structure analysis of noun phrases using Abstract Meaning Representation (AMR). AMR is a graph representation for the meaning of a sentence, in which noun phrases (NPs) are manually annotated with internal structure and semantic relations. We extract NPs from the AMR corpus and construct a data set of NP semantic structures. We also propose a transition-based algorithm which jointly identifies both the nodes in a semantic structure tree and semantic relations between them. Compared to the baseline, our method improves the performance of NP semantic structure analysis by 2.7 points, while further incorporating external dictionary boosts the performance by 7.1 points. language of the presentation: Japanese | ||
| 村上 和夫 | 1451108: M, 1回目発表 | 自然言語処理学 |
|
title:Enhancing review-based recommender
abstract:Recently, recommenders are introduced into many EC sites and these sites have review parts. Collaborative filtering approach is widely used for recommendation, but It consider only users'ratings. The purpose of this study is to enhance recommenders by introducing users' item reviews. I show mainly related works on this field and my future work. language of the presentation:Japanese | ||
| 札場 寛之 | 1551094: M, 1回目発表 | 知能コミュニケーション |
|
title: Attempts to Generate Source Code from Natural Language using Statistical Machine Translation
abstract: Programming language provides programmers many useful syntax. But if programmers don't know much about the programming language, it is difficult to write code with it. In order to solve this problem, we apply machine translation techniques to auto generate source code from natural language, which is considered to be more familiar to people. language of the presentation: Japanese | ||
| IN WOLDE LUEBKE ARNO | 1461016: D, 中間発表 | 加藤 博一，向川 康博，Christian Sandor，山本 豪志朗 |
|
title: Efficient Differential Rendering for Outdoor Augmented Reality abstract: By embedding virtual content into the real world, Augmented Reality (AR) provides an intuitive way to convey information.Notably, many application are located in outdoor environments, e.g. for presenting lost cultural heritage on site. For this purpose, mobile devices and in particular smart phones and tablets have become an ubiquitous platform as they integrate various sensors beneficial for AR. For visual plausible integration of virtual content given real world light conditions, differential rendering provides a framework, however, requires accurate measurement and simulation of light. Due to limited computational capabilities of mobile devices, efficient implementations of such processes are needed. Our research is concerned with developing such methods. In particular, we aim to exploit the low temporal variance of outdoor light by pre-capturing and -computing a set of common outdoor conditions in advance. Additionally, since AR scenes commonly exhibit low depth-complexity, we aim to investigate screen-space approximations of real and virtual scenes to efficiently compute effects of global illumination. language of the presentation: English | ||
| 神田 将輝 | 1551037: M, 1回目発表 | ロボティクス |
|
title: Development of low cost electric prosthesis which combines both aesthetic appeal and functionality
abstract: Currently, most transradial amputees use cosmetic prostheses that resemble a human arm. However, cosmetic prostheses are inconvenient in daily life because they do not have a holding function. As another option, transradial amputees can also use electric prostheses. However, they have not become popular because of their unnatural appearance, high price and heavy weight. In this work, we are developing a low cost electric prosthesis which combines both aesthetic appeal, through a design similar to a human arm, and functionality, by implementing a holding function. language of the presentation: Japanese 発表題目: 装飾性と機能性を兼ね備えた低コストな電動義手の開発 発表概要: 現在，国内のほとんどの前腕切断者は装飾義手を使用している． 装飾義手は手に近い外観を持っているが，把持機能がなく，日常生活では不便である． 他の選択肢として把持機能を持った筋電義手があるが，装飾義手よりも外観は劣り，価格が高く，重いため，普及が進んでいない． そこで，本研究では手に近い外観と把持機能を備えることにより，装飾性と機能性を両立する低コストの電動義手の開発を行う． | ||
| 中山 昴 | 1551069: M, 1回目発表 | 計算システムズ生物学 |
|
title: Motion correction algorithm for ECV Map in cardiac MRI abstract: Cardiac Magnetic Resonance Imaging is a well-established noninvasive imaging modality in clinical cardiology. Its accuracy in defining cardiac morphology and function are valid for diagnosing ischemic disease. Extracellular volume fraction (:ECV) indicate a value of the myocardial tissue quantitatively and ECV Map has been reported to have a high diagnostic performance by combining with other MRI image. Such as in order to create ECV Map, it is necessary to align the position of the heart between pre-contrast and post-contrast. In my research plan, I introduce the method to align their image. language of presentation: Japanese 発表題目:心臓MRIにおけるECV Map作成のための動き補正アルゴリズムの検討 発表概要:心臓MRIは非侵襲的な画像診断装置であり、心臓MRIで得られる機能画像と形態画像は心筋梗塞といった虚血性疾患の診断に有効である。心筋組織の線維化を定量的に示すECVをMappingしたECV Mapは他のMRI画像と比較することによって高い診断能を示す。しかし、ECV Mapを作成するためには造影剤投与前と造影剤投与後のMRI画像の心臓の位置を正確に補正する必要がある。本発表では、ECV Map作成のための造影剤投与前後の心臓の位置合わせ手法について述べる。 | ||
| 大内 里菜 | 1551021: M, 1回目発表 | 数理情報学 |
|
title: Analysis of interaction between dogs and humans using expertise of dog trainers.
abstract: Dogs are an animal species that has lived with humans from ancient times. It is said that dogs have gotten their own special skill to interact with humans over long periods of time. As one of such a skill, Nagasawa et al. showed that gaze interaction between dogs and humans promotes social bonding. However, there are few previous studies focused on key factors of the interaction between dogs and humans. In our study, we are planning to extract such factors by using gaze behavior of dog specialists that can reflect their knowledge and experience. Finally, I will mention a future plan of our study. language of the presentation: Japanese | ||
| 中川 豊 | 1351075: M, 1回目発表 | 情報基盤システム学 |
| JIARPAKDEE JIRAYUS | 1551124: M, 1回目発表 | ソフトウェア工学 |
|
Title: Understanding Unanswered Question through Human Aspect Features in StackOverflow
Abstract: Ever since internet became widely available, question and answer site has been used as a knowledge sharing service. Users ask what they want to know, then will be answered by the other users in a community. However, unfortunately, the questions do not always get answers. Eric Raymond claimed how user asks a question is important. Existing studies have presented approaches of the question quality by textual and community-based aspects. In this seminar, I will present my empirical study to investigate how human aspect features related to the possibility of getting answer. Language : English | ||
| 小芝 涼太 | 1551044: M, 1回目発表 | ユビキタスコンピューティングシステム |
|
title: A study on the user grouping using various techniques
abstract: By performing a user grouping to suit the context, we aims to make an opportunity of a more smooth communication in offices and schools. In the research on the user grouping, with the use of environmental sounds, radio waves, there is to do grouping of physical proximity. Also, by using the SNS and TV sound, there is to do grouping of logical proximity. In this presentation, after which we were introduced previous studies of, and to announce the research plans for the future. language of the presentation: Japanese 発表題目: 様々な手法を用いたユーザーグルーピングに関する研究 発表概要: コンテキストに合わせたユーザーグルーピングを行うことで、オフィスや学校等での【より 円滑なコミュニケーションのきっかけ作り】を目指している。ユーザーグルーピングに関する研究で は、環境音、電波を用いて物理的な近さからグルーピングを行うもの。またSNSやTV音声を用いること で、論理的な近さからグルーピングを行うものがある。本発表では、それらの先行研究を紹介した後 に、今後の研究予定について述べる。 | ||