ゼミナール発表

日時: 7月26日(水)3限 (13:30-15:00)


会場: L1

司会: 吉野 幸一郎
堀江 卓弘 1651203: M, 1回目発表 自然言語処理学 松本 裕治
 
LI MICHAEL WENTAO 1651217: M, 1回目発表 自然言語処理学 松本 裕治
title: Exploring the Meaning of Machine Translation Quality Estimation
abstract: The problem of how to evaluate machine translation (MT) output has led to the development of various automatic metrics, which in turned have spurred the rapid growth of MT over the past decade and a half. A related field to evaluation is that of quality estimation (QE): using only the source and translated output to predict how a human would judge a translation. However, the question of "what is a good translation?" by human standards remains a black box. Difficult-to-analyze sources of noise exist at almost every step of the pipeline, raising the question of what is actually being measured by QE, and what it means. My goal is to identify, quantify, and ideally develop a method to reduce some of these sources of noise.
language of the presentation: English
 
RUIZ CAMILLE MARIE 1651218: M, 1回目発表 自然言語処理学 松本 裕治☆
title: Loneliness in a Connected World: Analyzing Online Activity and Expressions on Real Life Relationships of Lonely Users
abstract: Although loneliness is a very familiar emotion, little is known about it. An aspect to explore is the prevalence of loneliness in the connected world that social media sites like Twitter provide. In light of this, this study investigates the Twitter data of users that have expressed loneliness to understand the phenomenon. Since our primary material are tweets, we developed various indices that can measure social activities reflected in online relationships and real life relationship solely through online Twitter data. Through these indices, the relations between social activity and loneliness were investigated. The results show that high lonely users seem to have low online activity, high positive expressions on real life relationships, and narrow ingroups.
language of the presentation: English
 
LU YUXUN 1651219: M, 1回目発表 自然言語処理学 松本 裕治
title: Common Structure Extraction from Raw Text of Scientific Literatures
abstract: As the increasing rate of papers being published in scientific field, it has became hard for individual researcher to fully keep up with the developments in their research field. Our research aims to extract common structure of scientific literatures in order to compare the differences and similarity between scientific literatures in a specific domain. Ideally, we hope such method of extracting common structure could generalize across the research domains. Currently, we are tackling with this task with two different approaches: (1) Topic Modeling; and (2) Template Extraction. However, both of them have some pros and cons and need to be modified for this task. In this presentation, we would introduce our basic motivation, the final goal we wish to archieve and compare the two models mentioned above.
language of the presentation: English
 
MUHAIMIN HADING 1551204: M, 2回目発表 自然言語処理学 松本 裕治, 松本 健一, 新保 仁, 進藤 裕之
title: Japanese Simplification for Non-Native Speaker
abstract: This introduces Japanese lexical simplification. Japanese lexical simplification is the task of replacing difficult words in a given sentences to produce new sentences with simple words without changing the original meaning of the sentences. We purpose a method of supervised regression learning to estimate the difficulty ordering of words with statistical features obtained from two types of Japanese corpus. For the similarity of the words we use a Japanese thesaurus, Japanese lexical simplification system, and dependency based word embeddings. We conduct two types of evaluation; (1) evaluation of the words difficulty is performed by comparing the difficulty ordering of the words, (2) evaluation of new simplified sentences by a human judge who is a Japanese learners and evaluation of kipping the original meeting by a human judge from Japanese native speaker.
language of the presentation: English
 

会場: L2

司会: 藤本 大介
TIAN ZHAOYIN 1651215: M, 1回目発表 計算機システムズ生物学 金谷 重彦
title: *** Development of a real-time detection algorithm of abnormal heart beat based on personal electrocardiogram ***
abstract: *** The classification of thousands of heartbeats is a challenging task even for experienced cardiologists, and moreover if the analysis is focused in subtle or infrequent arrhythmias. So, our goal is to develop a real-time detection algorithm of arrhythmias based on personal electrocardiogram, which can help people detect arrhythmias by themselves. Since Home-use or personal cardiac monitor can give more information in prevention stage and unrestricted monitoring cause it greater in interpatient variability, we decide to focus our research on Holter. A Holter monitor is a small, battery-powered and over 24 hours wearing medical device that measures your heartユs activity, such as rate and rhythm. Based on our real time demand, we decide to apply unsupervised machine learning method to classify the Electrocardiogram(ECG) signal. ***
language of the presentation: *** English ***
 
FAN JIAYING 1651216: M, 1回目発表 モバイルコンピューティング 伊藤 実
title: Consideration for a Within-Subject Study Comparing Effectiveness of Cameras with Different Field of Views in Remote Collaboration
abstract: As mobile communication devices and high definition cameras has been used extensively, remote collaboration can be supported in a lot of scenes in our daily life, like remote surgery, machine repairing, online learning. Except for voice, cameras also play an important role in view sharing in remote collaboration activities, especially with the emerging of panorama camera, more working environment information can be provided to distributed partners. Even though the cameras always being used in life recording and workspace sharing, their impact on remote collaboration is still unknown to us. To get a better understanding of effectiveness of different field of view in workspace sharing, while also in remote physical collaborative behavior, process, perception and result performance. We conduct a within-subject study contains two tasks and three technology methods, which tasks are setting to Lego assembly with detailed manual or not, the technical methods including voice, cameras and panorama cameras. We hope the findings from this study could offer design and research implications, underlining the importance of wide angle view of workspace in remote collaborations.
language of the presentation: Japanese
 
中川 豊 1351075: M, 1回目発表 情報基盤システム学 藤川 和利
title: Real time rainfall prediction by network of virtual rain sensor using radio wave attenuation
abstract: In recent years, attention has been paid to millimeter wave bands and quasi millimeter wave bands for the purpose of high speed wireless communication with a large capacity. 25 GHz radio communication equipment system has been constructed as a backup line also within our university campus and a method to virtually measure rainfall on the ground surface using the attenuation characteristics of radio waves due to precipitation was studied and I was able to reconfirm and validate its effectiveness. I will accumulate a lot of observation data in future and monitor the radio wave attenuation by other methods (such as the effect of rainfall of the BS antenna) to explore the construction of a system to estimate more accurate rainfall. I will further examine methods to combine with sensors other than rainfall and utilize sensor network to lead guerrilla heavy rain prediction.
language of the presentation: Japanese
発表題目:電波の減衰を利用した仮想降雨センサの構築とネットワークによるリアルタイム降雨予測
発表概要: 近年、大容量での高速無線通信を目的としてミリ波帯・準ミリ波帯が注目されている。当大学キャンパス内においてもバックアップ回線として25GHz の無線通信装置システムが構築されており、降水による電波の減衰する特性を利用した地表での降雨量を仮想的に測定するための手法が検討され、その有効性を再確認、検証することができた。 今後観測データを多く集積し、他の手法(BSアンテナの降雨による影響など)により電波減衰のモニタリングを行い、より精度の高い降雨量を推定するシステムの構築を探る。さらに雨量以外のセンサと組み合わせることやセンサネットワークを活用しゲリラ豪雨予測につながる手法を検討する。
 
佐藤 哲朗 1651052: M, 1回目発表 情報基盤システム学 藤川 和利
 
藤井 博晃 1551089: M, 1回目発表 数理情報学 池田 和司