この冬、といっても来週ですが、Microsoft Research AsiaからDeputy Managing DirectorであるDr. Feng Zhaoを含む3名の研究者が来日し、各地の大学と協力して講演会を行います。各場所によって若干フォーマットは異なりますが、大まかな会場と予定されている講演者は以下の通りです。詳しくは各リンク先をご参照ください。
Microsoftは今年、過去最大規模のプロダクトローンチイヤーを迎えています。Windows 8, Windows Phone 8, Office 2012, Windows Server 2012, などなど。今後のビジネス基盤を強化するために非常に重要な一年となっています。Microsoftの研究所であるMicrosoft Researchの第一のミッションは基礎研究への貢献であり、アカデミアのコミュニティに深くコミットし、世界各地の大学・研究機関と研究交流・人材交流を深めている一方で、Microsoft Researchで生まれた研究成果は少なからずこれらの製品へと技術移転され、多くの人々の生産性の向上に大きなインパクトを与えています。企業で基礎研究を行うことの醍醐味はこの点にあるといってもよいでしょう。
またここ数年、Microsoft Researchが推進しているCOREの連携研究プロジェクトやインターンシップを通じて、日本の大学の研究者の方々や学生の方々がMicrosoft Researchに訪れたり交流をおこなったりして、Microsoft Researchの研究者と連携を深め、ネットワークが広がっているのはうれしい限りです。
Microsoft Research Asiaというと、Multimedia, Natural Language Processing, Human Computer Interaction当の研究が日本の学生の方々になじみが深いかと思いますが、今回は自然言語処理の辻井先生のほか、以前Redmondで首席研究員としてNetworked Embedded Computing Groupを立ち上げ、現在は北京の研究所の副所長として、モバイル、ネットワーク関連のグループを統括しているFeng Zhaoと、北京のInternet Economics & Computational Advertising (IECA) groupを率い、ゲーム理論と機械学習をベースにしたオンライン広告に関する研究を行っているTie-Yan Liuが、彼らの最新の研究成果について話をいたします。
・2012年12月5日（水）午後２時40分～午後6時 東京大学 本郷キャンパス
・2012年12月6日（木）午後3時00分～午後5時 大阪大学 吹田キャンパス
・2012年12月7日（金）午前9時00分～午前12時 京都大学 吉田キャンパス
Dr. Feng Zhao, Assistant Managing Director, Microsoft Research Asia
Title: Mobile Sensing
The lofty vision of the wireless sensor network research when it started more than a decade ago was to blanket the planet with tiny, self-organizing smart dust. Each dust particle has a little bit of sensing, computation and communication, with some onboard energy reserve. When released in the ambience, the smart dust collaborates to sense and possibly act on the physical world and its inhabitants, for a variety of societal scale problems such as environment, energy, health, and mobility. Now, with the advent of the increasingly more capable sensors on widely available platforms such as cell phones and vehicles, the age of planet-scale sensor networks has finally arrived. This new generation of mobile sensing systems leverage storage and processing on both mobile devices and in the cloud. Furthermore, the ability to crowd-source the sensing and action with users in the loop presents new opportunities as well as raising issues of privacy and security. In this talk, I will first give a brief review of the major advances in sensor networks to date. The rest of the talk will be on mobile sensing, including sensing a person’s physiological state, mapping out noise in the environment, and understanding human mobility patterns for better urban planning.
Feng is an Assistant Managing Director at Microsoft Research Asia, responsible for the hardware, mobile and sensing, software analytics, systems and networking research areas. His own research has focused on wireless sensor networks, energy-efficient computing, and mobile systems. Prior to joining MSR-Asia in 2009, he was a Principal Researcher at MSR Redmond (2004-2009), and founded the Networked Embedded Computing Group that has designed and deployed sensor networks at several Microsoft datacenters for environmental monitoring and energy optimization. He was a Principal Scientist at Xerox PARC 1997-2004, and founded PARC’s sensor network effort.
Feng has championed the wireless sensor network and energy-efficient computing research at Microsoft. He was among the first to develop a suite of collaborative sensing and processing protocols for tracking problems using networked sensors, including the IDSQ algorithm. He authored or co-authored over 100 technical papers and books, including a book, Wireless Sensor Networks: An information processing approach, by Morgan Kaufmann. He was the founding Editor-In-Chief of ACM Transactions on Sensor Networks (2003-2010), and founded the ACM/IEEE IPSN conference. In 2008, he helped start a new workshop, HotPower, focusing on the emerging topic of sustainable computing.
Feng received a PhD in Computer Science from MIT, and taught at Ohio State University and Stanford University. An IEEE Fellow, Feng received a Sloan Research Fellowship (1994) and NSF and ONR Young Investigator Awards (1994, 1997).
Dr. Junichi Tsujii, Principal Researcher, Microsoft Research Asia
Title: Deep Parsing and Semantic Processing of Text: Linking Meanings with Text
Due to the ever increasing amount of text available in computer systems, natural language processing (NLP) is becoming one of crucial technologies which will contribute to shaping the future form of IT-based society. NLP is to be combined and merged with other IT technologies such as semantic web, data and text mining for big data, multi-media database, information retrieval, etc. Deep parsing plays a crucial role in linking meanings of text with “knowledge” or ‘information”. While NLP and NLU (Natural language Understanding) have been one of the most challenging research areas in Artificial Intelligence, recent progresses technologies in ontology construction, machine Learning, computational linguistics and their integration have opened up several new promising research agendas. Among them, I will focus on how deep parsing can be used in various applications such as semantic search, entity-based knowledge management, paraphrase recognition, multi-document summarization, etc. I will also talk about on-going projects at Microsoft Research Asia.
Prof. Junichi Tsujii, Principal Researcher of Microsoft Research Asia (MSRA). Before joining MSRA (May, 2011), he was Professor of Natural Language Processing in the Department of Computer Science, University of Tokyo and Professor of Text Mining in School of Computer Science, University of Manchester, U.K. . He remains to be scientific advisor of the UK National Centre for Text Mining (NaCTeM) as well as visiting professor of University of Manchester. He has worked since 1973 in Natural Language Processing, Question Answering, Text Mining and Machine Translation. He has received a number of awards such as the IBM Science Award (1989), SEYMF Visiting Professorship (2000), Daiwa-Adrian Prize (2004), IBM Faculty Award (2005) and Achievement Award of Japan Society for Artificial Intelligence (2008), Fellow of Information Processing Society Japan (2010) and the Medal of Honor with Purple Ribbons (2010). He was President of ACL (Association for Computational Linguistics, 2006) President of IAMT (International Association for Machine Translation (2002-2004), and President of AFNLP (Asian Association for Natural Language Processing, 2007). He is Permanent member of ICCL (International Committee for Computational Linguistics, 1992-), Member of the advisory board of Institute of Information Science, Academia Sinica in Taiwan (2011-), Member of SIG:MA (Scientific Innovation Group: Mentors and Advisors), Elsevier Inc. (2012-), etc.
Dr. Tie-Yan Liu, Research Manager, the Internet Economics & computational Advertising Group, Microsoft Research Asia
Title: Machine learning for computational advertising: challenges and opportunities
Online advertising is the revenue source of most online services. The key component of online advertising is an auction mechanism, which determines how the ad slots are allocated and how the advertisers are charged. Generalized second price auction (GSP) is the most popularly used auction mechanism in the industry. The wide adoption of GSP is partly because of some historical reason, and partly because it was proven to have quite nice theoretical properties. However, most theoretical analyses on GSP were obtained with strong assumptions, such as full information access and full rationality. However, such assumptions are largely different from the reality: there are a large number of advertisers competing with each other, and it is almost impossible for any of them to have full access to the information of his/her opponents so as to make the best decision. The real advertiser behaviors will hurt the theoretical properties of GSP, in particular, lead to a much lower revenue than expected. To solve the problem, we propose two solutions. First, instead of relying on assumptions, we learn an auction mechanism which can have much higher revenue than GSP on real data. Second, we learn an effective bid optimization tool to help advertisers improve their rationality so as to increase the competition in auction. These two approaches are highly non-trivial because conventional machine learning algorithms and theorems cannot be directly applied. We show how we tackle these challenges, and further discuss a promising research direction on its basis, which we call game-theoretic machine learning.
Tie-Yan Liu is the research manager of the Internet Economics & Computational Advertising (IECA) group of Microsoft Research Asia. His research interests include machine learning, game theory, computational advertising, and micro-economics. He is widely recognized as one of the best researchers in the field of learning to rank. He has authored two books, and tens of highly-cited papers in the related journals and conferences. He is the co-author of the best student paper for SIGIR (2008), and the most cited paper for the Journal of Visual Communication and Image Representation (2004-2006). He is a program committee co-chair of RIAO (2010), a demo/exhibit co-chair of KDD (2012), a track chair/area chair/senior PC of SIGIR (2008-2011), AIRS (2009-2011), WWW (2011), and IJCAI (2013). He is an associate editor of ACM Transactions on Information System (TOIS), an editorial board member of Information Retrieval Journal and ISRN Artificial Intelligence. He is a keynote speaker at PCM (2010) and CCIR (2011), a plenary panelist of KDD (2011), and a tutorial speaker at several conferences including SIGIR (2008, 2010, 2012), WWW (2008, 2009, 2011), and KDD (2012). Prior to joining Microsoft, he obtained his Ph.D. in electronic engineering from Tsinghua University. He is a senior member of the IEEE, ACM, and CCF. He is now an adjunct professor and Ph.D. supervisor of the Nankai University and the University of Science and Technology of China.