| 白石 磨貴男 | 1351058: M, 1回目発表 | 計算システムズ生物学 |
|
title: Simulation of Energy Metabolism with Chemiosmotic Coupling abstract: There are many hypothesis about the origin of life. Above of all, "Deep Sea Vents Hypothesis" explains each stage of early evolution using knowledge of exiting landform and microbial. In a paper "Early bioenergetic evolution", Filipa L. Sousa et al., main adovocators of the hypothesis, argued that the antiquity of energy metabollism is chemiosmotic coupling which had a leaky membrane and promiscuous ATPases. However, it is still unclear that the early chemiosmotic coupling system behaves as the hypothesis predicts. My goal is to simulate this system including an ion gradient. In this presentation I will introduce the paper of Filipa et al. and explain my research plan. language of the presentation: Japanese | ||
| 伊藤 冬樹 | 1351009: M, 1回目発表 | 知能システム制御 |
|
title: An Approach to Feedforward Learning Control for Nonlinear Systems. abstract: A learning approach to feedforward controller on 2 degree of freedom control system is known for an effective approach. In resent years, feedback error learning and adaptive control for non-linear systems have been especially proposed. This presentation outlines a paper of neural-adaptive control. This shows that digital weight update causes the neural network to estimate a discrete-time model of the system, assuming that state feedback is still applied in continuous time and the update frequency and using the extra computational time to increase the size/accuracy of the neural network result in better performance. Experimental results achieving link tracking of a two-link flexible-joint robot verify the improved performance. language of the presentation: Japanese | ||
| 宇田川 大二郎 | 1351012: M, 1回目発表 | 知能システム制御 |
|
title: Model Predictive Control for Cooling Process of Hot-Rolled Steel Strip
abstract: Recently, it is required to improve a quality of a hot-rolled steel strip. The quality of the strip depends on its thermal transition in a cooling process of manufacturing plant. Thus the control of cooling process attracts many researchersユ attention. Model Predictive Control (MPC) is often used to control process plant. Since MPC has to calculate optimal inputs in a particular time step, in order to implement MPC in real, large-scale and relatively fast system, it is necessary to reduce its computational effort. In this presentation, a paper, which presents a method to reduce computational efforts, is introduced. In the paper, Distributed Model Predictive control (DMPC) is used to reduce computational effort. Moreover, neighborhood optimization is also used to deal with a problem that solutions of distributed optimization problem are different from ones of general optimization problem. At the end of this presentation, my future works are introduced. language of the presentation: Japanese | ||
| 長谷川 嵩大 | 1351087: M, 1回目発表 | 知能システム制御 |
|
title:A study of collaborative decision making using user's bioinformation and robot's sensor information
abstract: In this study, we aim to develop an interface that uses human's EMG signals as input. When we operate a robot system using such an interface, we need to consider uncertainties caused by human's decision making and noise of EMG measurement. To cope with these uncertainties, we propose an interface that uses both human's EMG and robot's sensor signals based on the approach of probabilistic collaborative-decision-making. In this presentation, we also introduce a paper related to the probabilistic collaborative-decision-making. language of the presentation: Japanese | ||
| 仲田 昌司 | 1351078: M, 1回目発表 | ロボティクス |
|
title: Using of color information in object recognition
abstract: In this presentation, I describe survey on the use of color information in image processing and progress of research, outlook for the future.In the object recognition, we use gray scale image primarily. If we attempt to use color values, a problem is occur about color constancy.And I introduce sparse coding. language of the presentation: Japanese 発表題目: 物体認識における色情報の利用 発表概要: 本発表では，画像処理における色情報の活用についてのサーベイおよび研究の進捗今後の展望について述べる．物体認識では，主にグレースケール画像を利用している一方で，色情報を利用して物体認識を行おうとすると色恒常性の問題がある．脳の一次視覚野の数学的モデルであるスパースコーディングを紹介する． | ||
| 佐原 壮海 | 1351053: M, 1回目発表 | インターネット工学 |
|
title: Structure of Rate Adaptation
abstract: In recent years, the transmission rate of Wireless LAN (Local Area Network) is increasing due to transmitting much more data. IEEE 802.11, which is a standard of Wireless LAN, specifies multiple transmission rates but there is no particular algorithm that selects those transmission rates. Currently, Rate Adaptation (RA) is being adapted as a technique to select transmission rate according to the quality of a wireless channel. In this presentation, I will describe the existing RA algorithms and discuss the features and general issues. Finally, I will also discuss how to approach these problems. language of the presentation: Japanese 発表題目: Rate Adaptation の仕組み 発表概要: 近年、無線 LAN (Local Area Network) の伝送レートは高速化している。しかし、無線 LAN の規格である IEEE 802.11 では、複数の伝送レートは規定するが、それらの伝送レートを選択するアルゴリズムは標準では規定していない。そのため、無線チャネルの品質に応じて適切な伝送レートを選択する Rate Adaptation (RA) の技術が研究されている。本発表では、既存の RA の手法と特徴、課題を示し、自身の研究方針について述べる。 | ||
| NAKASAN CHAWANAT | 1351117: M, 1回目発表 | ソフトウェア設計学 |
|
title: GFARM V2: A grid file system that supports high-performance distributed and parallel data computing
abstract: This presentation outlines the Gfarm distributed storage system, which was created to support global data-intensive computing. It is designed with parallel computing and job scheduling in mind. We will briefly explore the general concepts of Gfarm, then we introduce the presenter's research topic. language of the presentation: English | ||
| CHINDAMAIKUL KHANA | 1351118: M, 1回目発表 | ソフトウェア設計学 |
|
title: *** Constructing Assurance Case using Information From an Issue Tracking System ***
abstract: *** We address the problem of constructing an assurance case by presenting an approach to find some issues in issue tracking system that can be information for constructing an assurance case. We take advantage of document retrieval and topic modeling techniques to elicit relevant information which can be used as materials for constructing assurance cases. This paper gives an overview of an approach and reports the result of some preliminary experiment. The results suggest that the proposed approach could be effective in terms of reducing time and cost for constructing an assurance case. *** language of the presentation: *** English *** | ||
| 小林 靖幸 | 1351042: M, 1回目発表 | 計算メカニズム学 |
|
title: [Paper introduction] A Provably Secure and Efficient Countermeasure Against Timing Attacks
abstract: Timing attack is a side channel attacks which the attacker obtains the secret information in the cryptsystem by analyzing execution time. Countermeasures against timing attacks have been proposed input blinding and constant-time implementation. However, in the former, A recent result shows that input blinding may leak the secret information. in the later, it is not widely used in practical implementations of cryptosystems. In previous research, we will introduce a paper "A Provably Secure and Efficient Countermeasure Against Timing Attacks" by Köpf et al. that propose a countermeasure against timing attacks using quantitative information flow and Guessing Entropy. Also I will present my research plan. language of the presentation: Japanese 発表題目: [論文紹介] A Provably Secure and Efficient Countermeasure Against Timing Attacks 発表概要: サイドチャネル攻撃の1つであるタイミング攻撃は，暗号システムの実行時間を元に秘密情報を得る攻撃である．対策として，input blindingとconstant-time implementationがあるが前者は攻撃手法が提案され，後者は実装が困難という問題がある．そこでまず従来研究としてKöpfらの論文“A Provably Secure and Efficient Countermeasure Against Timing Attacks”を紹介する．この論文は量的情報流と推測エントロピーを用いた，タイミング攻撃への対応策を述べている． 最後に今後の研究方針について述べる． | ||
| 落合 誠治 | 1351024: M, 1回目発表 | インタラクティブメディア設計学 |
|
title: A Study of Arm Extending Illusion by Visual and Vibrotactile Stimuli abstract: Direct manipulation of virtual objects is often used in Augmented Reality because tangible user interface is considered intuitive.However, the manipulation has physical limitations, for example, the user can interact with only the objects that are in the range of the arms reach. In this research, we have tried to realise an arm extending illusion by visual and vibrotactile stimuli for reaching to far-situated objects. Firstly, only vibrotactile stimulus to the tendon of the biceps was used, then cross-modal effect by visual and vibrotactile stimuli was conducted out with test users. In this presentation, we report the prototype system which gives the user the stimuli and also present the result of the brief experiments using the system. language of the presentation: Japanese 発表題目: 視触覚刺激による腕部伸長感覚提示の検討 発表概要: 拡張現実感技術では現実空間に仮想物体の重畳表示が行え、それらとのインタラクション方法が検討されている。その一つの方法として直接手を用いたインタラクションが考えられる。自身の手を用いることで自然な操作が可能だが一方で操作可能な範囲は手の届く範囲に限られるという問題が発生する。そこで本研究では 遠方物体へのインタラクションにおいても手を用いて操作を可能とするような 腕の伸長感覚の提示方法について検討する。上腕二頭筋の腱への振動刺激による運動感覚の生起、視覚によるクロスモーダル効果の二つを組み合わせることで腕の伸長感覚を発生させる。本発表ではその進捗報告として、試作したモーターを利用した振動刺激装置とCGモデルを利用した視覚刺激動画、そしてこれについて行った予備実験について報告する。 | ||
| 片桐 敬太 | 1351030: M, 1回目発表 | 視覚情報メディア |
|
title: Depth Synthesis and Local Warps for Plausible Image-based Navigation
abstract: Image-based rendering (IBR) is a technique to synthesize an arbitrary viewpoint image from multiple input images capturing a scene, and recent research efforts have been dedicated to its variant that uses three-dimensional (3D) geometry of the scene obtained by using a multi-view stereo (MVS) technique. This approach transforms the input images to the view from the specified viewpoint based on the geometry and reduces the number of input images required for synthesizing a plausible image. However, MVS often fails in reconstructing object with complex geometry, resulting in disturbing visual artifacts. The method proposed in this paper represents the 3D geometry by depth values for each input images and estimates appropriate depth values in regions without depth values due to failure of 3D reconstruction. The method reduces the visual artifacts due to failure of 3D reconstruction by transforming the input images based on the estimated depth values. In this presentation, I describe the method in detail and its limitation, as well as my research direction. language of the presentation: Japanese 発表題目: 違和感がないイメージベースナビゲーションのための奥行き合成とローカルワープ 発表概要: Image-based rendering (IBR)はあるシーンを撮影した複数の画像を入力として、任意の視点の画像を生成する自由視点画像生成技術のひとつであり、近年、Multi-view stereo (MVS)などを用いて復元されるシーンの３次元形状に基づいて入力画像を変形することにより、より少数の入力画像から視覚的な違和感が少ない任意の視点からの画像を生成する手法が提案されている。しかし、MVSなどでは複雑な形状を復元できないため、生成された画像が歪むという問題があった。本論文では、MVSで復元される３次元形状を各入力画像の奥行き値で表現し、３次元形状が復元できずに奥行き値が得られなかった領域に対して、その周辺の奥行き値などに基づいて奥行き値を推定することにより、歪みの少ない画像を生成するIBR手法を提案する。本発表ではその手法と課題を説明して、自身の研究方針について述べる。 | ||
| 金丸 智史 | 1351031: M, 1回目発表 | 自然言語処理学 |
|
title: Technique of Utilizing Japanese-English Resource for Sentiment Classification
abstract: Sentiment classification is the task of classifying sentiment of a given text as positive or negative.A classifier constructed by supervised learning is usually used for this task.However, there is insufficient training data with accuracy labels in Japanese.In this presentation, we describe the technique of using labeled English data which exists a lot and unlabeled Japanese data for constructing a classifier. language of the presentation: Japanese 発表題目: 感情極性分類のための日英資源の活用手法 発表概要: 感情極性分類とは、与えられたテキストの感情極性を特定するタスクである。教師あり学習を使用した分類器を用いてこのタスクは行われている。しかし、日本語の正解ラベル付きの学習データが不足している。そこで、豊富に存在する英語の正解ラベル付き学習データと正解ラベルなし日本語の学習データを組合せて、分類器構築に活用する手法を述べる。 | ||
| 波多腰 優斗 | 1351088: M, 1回目発表 | 知能コミュニケーション |
|
title: Construction of Tree-to-String translation system using the rules of pre-ordering
abstract: Statistical machine translation can learn translation rules from parallel corpora automatically. However, it has the problem that the amount of rules will be enormous. There has been considarable interest in pre-ordering, the method of ordering source language into the target language order by some simple rules before machine translation. In this reserch, we propose the method of using rules of pre-ordering in statistical machine translation in order to construct translation system with high accuracy and a small number of rules. language of the presentation: Japanese 発表題目: 事前並べ替え手法のルールを利用したTree-to-String 翻訳システムの構築 発表概要: 統計的機械翻訳は対訳コーパスから翻訳規則を自動的に学習できるが，膨大な量の規則が学習されてしまうという問題点がある．近年は，幾つかの簡単な規則によって，原言語を目的言語の語順に並べ替えてから統計的機械翻訳を行う事前並べ替えの手法が注目されている．そこで，本研究では少ない翻訳規則で高精度な翻訳を行うシステムを構築するために，事前並べ替え手法のルールを統計的機械翻訳の枠組みの中で利用する方法を提案する． | ||
| 蓑田 由花里 | 1351105: M, 1回目発表 | 知能コミュニケーション |
|
title: Evaluation of Social Skill Training through Observation of Brain Activation
abstract: Nonverbal communication skills are important in the daily life. However, people not good at nonverbal communication exist, and people with autism spectrum disorders are extreme example in them. For these people, social skills training (SST) can be used to train nonverbal communication skills. We have not yet established evaluation method for SST, but if we can develop an objective evaluation measure for SST, we can reveal characteristics of training and offer more efficient training for users. In this study, I will research brain activation during SST and introduce the way of objective and phisiological evaluation of SST. language of the presentation:Japanese 発表題目:脳賦活の観察によるソーシャルスキルトレーニングの評価 発表概要: コミュニケーション能力の中でも、非言語コミュニケーション能力は日常生活において重要だとされている。しかし、 非言語コミュニケーションを苦手とする人々も存在し、その極端な例が自閉症スペクトラムの人々である。このような人達のために、 非言語コミュニケーション能力を向上させるソーシャルスキルトレーニング（SST）という療法が存在する。 SSTの評価手法は未だ確立されていないが、SSTを客観的に評価することでトレーニング毎の特徴が分かり、より効率的なトレーニングを利用者に提供できる。 そこで、本研究ではSST遂行時の脳の賦活状況を調べ、客観的かつ生理学的にSSTを評価する手法を提案する。 | ||