| TRONO EDGAR MARKO SARMIENTO | 1461021: D, 中間発表 | 安本 慶一，笠原 正治，荒川 豊，諏訪 博彦，藤本 まなと |
|
title: Generating Disaster Area Pedestrian Maps using Distributed Computing across a Delay Tolerant Network
abstract: During disasters, maps of the affected area are needed to guide evacuees to safe areas or to lead responders to points-of-interest. However, the generation of digital disaster maps is challenging because it requires heavy processing, such as GPS trace aggregation and event detection from images, and the disaster may have damaged the communication network infrastructure, rendering Cloud-based mapping services inaccessible. The most available form of computing in such Cloud-less environments are mobile devices, but with limited processing speed and energy, these are not suited for mapping. In this study, we propose and evaluate system that leverages distributed computing across a Delay Tolerant Network to generate pedestrian maps. To realize our system, we address the following technical challenges: (1) how to collect data from the disaster area that can be used to generate a map, (2) how to process the collected data to generate maps in a Cloud-less disaster environment, and (3) how to propagate data across the system without a continuous end-to-end network. For the first challenge, we develop an Android application called DTN MapEx, which enables its users (i.e. pedestrian evacuees and on-foot responders) to use their mobile devices to collect disaster area data, specifically GPS traces. For the second challenge, because the Cloud is inaccessible and mobile devices do not have enough computing resources, we distribute data processing tasks to the key component of the system: Computing Nodes (CNs), which are pre-deployed, stationary kiosks with more computing power than mobile devices. The collected data are sent to the CNs, which run a pedestrian map inference program to generate a weighted directed graph that shows pedestrian paths and the average walking speeds along the paths. The generated maps are then sent to querying mobile devices. For the third challenge, we use Delay Tolerant Networking to propagate the collected data and generated maps across the system. We evaluate our system through experimentation and simulations, and posit that our approach can create maps faster than when data is processed locally in mobile devices. language of the presentation: English | ||
| 河村 一輝 | 1451041: M, 2回目発表 | 安本 慶一，中村 哲，荒川 豊，諏訪 博彦，藤本 まなと |
|
title: Customer Recommendation to Support Real Estate Salespersons for a Restaurant
abstract: In this study, we propose a method to support the real estate sales for the restaurant.Flow of customers leading to conclusion of a contract is as follows. ��They look for real estate in the search site. ��They view to the favorite properties. ��They apply for heir favorite propertie. ��They negotiate the property owner and reach the conclusion of constract. Flow�、���, �□��� are very difficult, so salesman urge to the next step by phone. In order to improve sales performance, this study recmmend the customer who is easy to apply for the property. language of the presentation: *** Japanese *** 発表題目: 飲食店向け不動産営業を支援する 申し込み顧客推薦手法の提案 発表概要: 本研究では, 飲食店向けの不動産営業を支援する手法を提案する. 飲食店起業を目指す顧客が, 物件の成約に至る流れは以下の通りである. �仝〆�サイトで不動産を探す. �⊆尊櫃貌盡�を行い気に入った物件を閲覧する. �Ｆ盡�に行った物件の内気に入った物件を申し込みする. �な�件の貸主と交渉し, 成約に至る. �、��△慮〆�から内見に移ることや, �□���の内見から申し込みに移ることは難しいため, 不動産会社の営業マンが電話をかけて次のステップに促す. 従来では電話営業を行う際, 営業マンは長年の感と経験によって, 数ある顧客リストからより成約に至りやすい顧客を選定し, 電話をかけていた. しかし, 新人の営業マンでは, どの顧客に電話をかけるべきかのノウハウがたまっておらず, どの顧客にかけたらいいのか分からないといった問題がある. この問題を解決するために, 本研究では機械学習を用いて, 申込みをしやすい顧客を推定し，営業マンに電話をかけるよう推薦する. | ||
| 藤澤 和輝 | 1451089: M, 2回目発表 | 安本 慶一，中村 哲，荒川 豊，諏訪 博彦，藤本 まなと |
|
title: Study on Automatic Content Curation System for Multiple Live Sport Video Streams
abstract: Our goal is to create personalized and high-presence multi-channel contents for a sport game through realtime content curation from various media streams captured/created by spectators. We use the live TV broadcast as a training data and construct a machine learning-based model to automatically conduct curation from multiple videos which spectators captured from different angles and zoom levels. As inputs for constructing a model, we use meta data such as image feature data (e.g., a pitcher is on the screen) in each fixed interval of baseball videos and game progress data (e.g., the inning number and the batting order). Output is the camera ID (among multiple cameras of spectators) at each point of time. For evaluation, we targeted Spring-Selection high-school baseball games. As training data, we used image features, game progress data, and the camera position at each point of time in the TV broadcast. Random Forests algorithm is used for learning. We used videos of a baseball game captured from 7 different points in Hanshin Koshien Stadium with handy video cameras and generated sample data set by dividing the videos to fixed interval segments. As a result,our method predicted the camera switching timings with accuracy (F-measure) of 72.53% on weighted average for the base camera work and 92.1% for the fixed camera work. language of the presentation: Japanese タイトル: 複数カメラで撮影したスポーツ映像ストリームの実時間自動キュレーションシステムの提案 概要: TV局の放送するスポーツ中継の映像を教師データとして，一般のスポーツの観 戦者が撮影した複数 映像ストリームのキュレーションを機械学習により自動で行うシステムを提案する． 入力データは各映像ストリーム中の各区間データに対する画像特徴情報と， イニング数や打順等の 試合進行情報といったメタ情報を用いる． 出力データは，各時刻における，TVで放送される映像に近い映像を撮影して いる一般の観戦者のカ メラのID，すなわちカメラの切替えタイミングである． 評価実験として，高校野球を題材に，阪神甲子園球場で撮影した7台のカメラ 映像に付与された画像 特徴情報と試合進行情報，TV局の放送映像を対象としてRandom Forestsを用 いて切替えタイミング 推定モデルを構築した． 提案手法を用いて，1〜3イニング目のカメラの切替えタイミングを学習させ たところ，カメラマン の独自性が反映されない決められたルールに則ったベースカメラ間の切替え については加重平均で 72.53%，固定カメラワークの切替えについては92.1%のF値を得た． | ||
| 田中 雄希 | 1451071: M, 2回目発表 | 金谷 重彦，池田 和司，MD.ALTAF-UL-AMIN，杉浦 忠男，佐藤 哲大 |
|
title: title: Analysis of efficacy from herbal medicine recipe abstract:The many countries use herbal medicine which utilizes medical plant. For example Japan (kampo) and Indonesia (Jamu). Estimation of efficacy from their recipe is important for standardrzation of these medicine. Moreover, Knapsack is metabollite database including herbal medicines and compounds in their plant.By combing information of such prediction models and metabolite database will allow us to find compounds which have not discovered in plant of the present recipes.In this presentation , I will introduce the result of prediction and analysis from plants and compounds in Jamu recipe by using Random Forest and Bayesian filter. language of the presentation: Japanese | ||
| 矢野 完人 | 1451114: M, 2回目発表 | 金谷 重彦，小笠原 司，MD.ALTAF-UL-AMIN，杉浦 忠男 |
|
title: Three dimensional structural analysis of dendritic spines
abstract: Recently, research about the brain have been actively performed. However, we don't know the brain very well. So we need to verify about the brain very well. There are a variety of techniques to analyze the brain. We have analyzed the brains by using two-photon excited fluorescence microscopy. And, I have focused on dendritic spines. I have studied a method of analyzation in the morphological features dendritic spines. language of the presentation: Japanese | ||
| LIDWINA AYU ANDARINI | 1451125: M, 2回目発表 | 金谷 重彦，小笠原 司，MD.ALTAF-UL-AMIN，杉浦 忠男，佐藤 哲大 |
|
title: Distant Respiration and Heartrate measurement with Doppler Radar abstract: The use of Doppler radar has proven to be one of the method of measuring heart rate and respiration rate without having to make contact with the subject. In this research the effect of movement and posture is being evaluated in correlation with the accuracy of the measurement. Here we are using 10.5 GHz doppler radar, while measuring the subject in resting positions. language of the presentation: English | ||
| 山崎 亘 | 1451116: M, 2回目発表 | 小笠原 司，横矢 直和，高松 淳 |
|
title: Acquisition of Semantic Objects Based on Classifying Motion of Human Hands
abstract: Service robots need to break up a simple instruction into detailed sub-tasks. It is important that robots have the knowledge of the relation between an object and manipulation of it because a sub-task is composed of a set of them. The purpose of my research is acquisition of semantic objects based on classifying the motion of human hands which is acquired from egocentric RGB-D sensor. language of the presentation: Japanese 発表題目: 人間の手動作分類によるセマンティックオブジェクトの獲得 発表概要: サービスロボットが簡便な指示によってタスクを実行する際，自身で詳細な動作，すなわちサブタスクに分解する必要がある．サブタスクは物体とそれに対する動作からなっているため，それらの関係性を知ることは非常に重要である．本研究では，一人称視点RGB-Dセンサを用いて物体に対する人間の手動作を取得し，その動作を分類して物体に対応付けることで，意味付けされた物体，すなわちセマンティックオブジェクトの獲得を目的とする． | ||