Computer Science Cafe - MSR Forum with top researchers
10:00 - 10:10 オープニングスピーチ 黒橋禎夫教授（京大情報学研究科）
11:00 - 12:00 京都のコンピュータサイエンス（京大関連の若手研究者・企業家の発表・質疑応答）
12:00 - 12:45 ポスターランチ（京大情報学の研究者によるポスター発表，軽食付き）
Mike Schroeder (Assistant Managing Director, MSR Silicon Valley)
Baining Guo (Assistant Managing Director, MSR Asia)
Dan Reed (Corporate Vice President, Extreme Computing Group)
Mike Schroeder (Assistant managing director, MSR Silicon Valley)
Baining Guo (Assistant Managing Director, MSR Asia)
Dan Reed (Vice President, Extreme Computing Group)
Shohei Nobuhara (Assistant Professor, Graduate School of Informatics)
Title: 3D Human Sensing using 3D Video
The development of 3D video in recent years realizes 3D
surface capturing of human in motion as is. In this presentation we
introduce 3D human sensing algorithms based on 3D video. Since 3D video
capturing does not require the object to attach special markers or to
wear a special costume, we can capture the original information such as
body motion or viewing directions without any disturbance caused by the
sensing system itself.
Satoshi Nakamura (Lecturer, Graduate School of Informatics)
Title: Reranking interfaces for bridging between users and search services
Nowadays, Web search services have become common tools to obtain information on the Web.
However, it is still not easy for users to find target information or required objects because of
the poor interaction between users and search services. In this talk, we have
proposed and realized a method for enabling users to rerank search results according to their search
intentions by direct manipulation. In addition, we have proposed several interfaces which
encourage users to interact with search results and enable them to compare something, to search
video clips and to rerank based on social bookmarks.
Naohisa Sakamoto (Assistant Professor, Center for the Promotion of Excellence in Higher Education)
Title: High quality particle-based volume rendering for large-scale irregular volume dataset
We describe an approach for sorting-free volume rendering of an
irregular volume dataset. Although the importance of irregular volume
rendering is widely recognized, scalability has been thought to be
difficult to realize since the large-scale irregular volume dataset
cannot fit into the memory space of the visualization node. A
ray-casting technique requires cell adjacency information in addition
to the volume dataset. A projection technique requires a large memory
space in order to calculate the visibility sorting at each viewing
point. To solve the problem, we regard the brightness equation as an
expected value of the luminosity of a sampling point along a viewing
ray, and propose a sorting-free approach that simply controls the
fragment rendering by using the evaluated opacity value to calculate a
rendered image. We applied our technique to large-scale irregular
volumes and confirmed that the splatting shows the best performance.
Akihiro Hatanaka (CTO, Questetra Inc., alumnus of Graduate School of Informatics)
Title: Smart start-up of SaaS Business -- in the case of Questetra BPM
alumni of graduate school of Informatics, Kyoto University.
We are developing and selling a BPM Software named "Questetra BPM
Suite" and our mission is "to optimize the world's human task-centric
When we launch Questetra BPM Suite, we distribute the software solely
as a download package.
However, since September 2009, we have been offering the software as a
The development of cloud computing services in recent years has been
lowering risk and cost of launching a SaaS even for start-ups.
In the presentation, we will give an overview about cloud computing
services in Japan and will introduce how we built the SaaS.
Toru Takahashi (Assistant Professor, Graduate School of Informatics)
Title: Simultaneous Speech Recognition System implemented on Humanoid Robot HRP-2
We developed simultaneous speech recognition system and implemented on
a humanoid robot. This system can recognize speech signals uttered by more than two
talkers. We often meet such a situation in real environment. Our system is constructed
with array of microphones, sound source localization, sound source separation,
acoustic feature extraction, automatic missing-feature mask generation and automatic speech
recognition based on missing-feature-theory. We conducted simultaneous speech
recognition experiment by three talkers. Our system was evaluated by work correct ratio
and compared with automatic speech recognition system without missing feature mask.
We show that our system outperform conventional system.
Megumi Kaneko (Graduate School of Informatics, Kyoto University)
Title: Radio Resource Allocation Algorithms for Wireless Multi-Carrier Relayed Systems
We focus on the issue of radio resource allocation algorithms for
downlink transmissions in a cellular system based on Orthogonal
Frequency Division Multiplexing (OFDM) technology. Recently,
increasing attention has been turned to cellular systems with relays,
but only a few algorithms have been designed for an OFDMA-based relay
system due to its inherent complexity. In this work, we propose low
complexity algorithms that operate in a relay-aided based station
centralized manner, thereby reducing the amount of required channel
state information. Simulation results show that our algorithms achieve
a very good throughput/outage trade-off.
Aki-Hiro Sato and Takeshi Nakamoto* (Graduate School of Informatics, Kyoto University)
Title:Comparative analysis of participants' synchrony and similarity of collective behaviour in the foreign exchange market
Vast amounts of high resolution data on human activities
are available due to development and spread of Information and
Communication Technology. In this poster we will show several
results of empirical analysis on the foreign exchange market with
the vast amounts of data on its quotes and trades. Findings and
conclusion are as follows.
1. Shares of currencies and currency pairs during each observation
period temporally vary and sometimes drastically changed.
2. There exists a positive correlation between synchronous indices
which measure impulsive concentration of participants' activities
over the market and a change of their shares.
It is concluded that both market participants' synchrony and
shares of currencies and currency pairs may influence trend of
Tomohide Shibata (Graduate School of Informatics, Kyoto University)
Title: Web Information Organization using Keyword Distillation Based Clustering
We propose a system that conducts search result clustering for several
thousands of Web pages, and elaborates cluster labels through keyword
distillation. Keyword distillation is a method that properly handles
spelling variations, transliterations, synonyms, and inclusion
relations, using linguistic resources. The system takes advantage of
a search engine infrastructure and grid computing environment.
Experimental results show that the system correctly merged synonymous
keywords and is useful for finding topics hidden in the lower-ranked
pages in a search result.
Yasuyuki Sumi (Graduate School of Informatics, Kyoto University)
Title: PhotoChat: Communication Support System based on Sharing Photos and Notes
I will show PhotoChat, a system that facilitates communication among
users who want to share experiences by enabling them to share photos
and notes. PhotoChat is designed to be used as a digital camera and
to run on mobile PCs with a camera module. PhotoChat users can
comment on the shared photos with a pen interface. The data, i.e.,
photos and comments, are distributed among PhotoChat users in real
time to enable them to learn others' interests and to chat easily.
Xinpeng Zhang and Yasuhito Asano (Graduate School of Informatics, Kyoto University)
Title: Discovery of "Enishi"
"Enishi" (縁) means relationship in Japan. Two kinds of relations
exist, explicit relations, e.g. a friend,
and implicit relations, e.g. a friend of a friend. Both kinds of relations
have important roles in the real world.
In this research, we propose a method for analyzing relationships between
objects on the link structure of Wikipedia using a generalized maximum flow
We also propose a tool for deeply understanding relationships based on the
Masaaki Nagahara (Graduate School of Informatics, Kyoto University)
Title: Activity report of Informatics Young Colloquium in Kyoto University
Informatics young colloquium is a luncheon seminar
on informatics held in Kyoto university.
This is held 2 or 3 times a month.
The topics are choosen from a large selection of
reseach areas such as computer vision,
machine learning, speech recognition,
control theory, signal processing,
numerical analysis, and so on.
In this presentation, we show the report
of this activity.