• Natural User Interaction: Exploring Future Computer Interfaces Today
  Session Chair: Frank Soong, Microsoft Research Asia
  
  We live in an era in which the traditional Windows, Icon, Menu, Pointing Device (WIMP) desktop paradigm, as well as our interaction devices—including the keyboard, mouse, and monitor—are becoming vastly inadequate. This is especially true given the rapidly evolving ecology of computing devices coupled with the growing information store and computational capabilities offered by the cloud. In fact, the bottleneck of how useful computers can be in our everyday lives is now largely bounded by the efficacy of our ability to interact seamlessly with them. Recent efforts to grow the bandwidth of human-computer interaction have culminated in an explosion of Natural User Interface technologies.
  
  In this workshop, researchers from Microsoft Research as well as our academic counterparts will present a broad survey of work inventing various pieces that will play key roles in the future of computing interfaces. These include advances in speech, gesture, augmented reality applications, Natural Language Processing, and even further out modalities such as brain-computer interfaces. 

A Universal Approach to Synthesizing High Quality Speech and Photo-Real Talking

Frank Soong, Microsoft Research Asia
In this talk, a universal algorithm is presented for synthesizing high quality speech and photo-realistic talking head. A database is first collected first to train a statistical, generative Hidden Markov Model (HMM), where trajectory of speech sound or articulator movement is trained and characterized with corresponding statistical HMM parameters. Thus trained HMM can then be used to predict underlying acoustic or articulator trajectories for given text or speech signals. The predicted trajectory by HMM are in turn to be used for guiding a sample based unit selection based synthesis. Appropriate samples (N nearest neighbors) in the original database are selected to for a “sausage” network and Viterbi algorithm is used to optimize a path search in the network to find the final high quality, speech/articulator movement output. Different training criteria are compared and demos for natural user interfacing will be presented.

Toward Music Listening Interfaces in the Future
Masataka Goto, National Institute of Advanced Industrial Science and Technology
In this talk, Active Music Listening interfaces and speech-recognition interfaces for music information retrieval are presented to demonstrate how end-users can benefit from understanding music and speech interaction technologies. First, Active Music Listening aims at allowing the user to better understand music and to actively influence the listening experience. Secondly, our hands-free music information retrieval systems employ two different speech-recognition interfaces, Speech Completion and Speech Spotter, which exploit intentionally controlled nonverbal speech information in original ways. Even if a user only remembers part of the title and the artist’s name or is talking to another person, the user can retrieve music. Our interfaces thus enrich end-users' music listening experiences and open up new ways of music listening in the future.

Be What’s Next – Context-aware Intelligent Companions (CAIC)
Chin-Yew Lin, Microsoft Research Asia
In Dr. Vannevar Bush’s seminal “As We May Think” article published in The Atlantic in July 1945, he laid out ideas about what scientists should do next to continue their effective partnership in response to a common cause during the Second World War. He constantly brought up the challenge of how people can freely record, store, modify, and consult the inherited knowledge acquired over the ages by humans. In particular, he mentioned a memex device which a user can stores all his books, records, and communications, and is constructed in such a way that instant access and manipulation of these data is possible.

Today, thanks to the invention of the World Wide Web, the rapid digitalization of all kinds of information, and the popularity of social media, we live in a world which Dr. Bush envisioned 65 years ago unfolding in front of our own eyes. Companies such as Microsoft, Google, and Yahoo! give their users huge online storage for free to keep their personal data. Digital cameras and HD camcorders are commonplaces. Data recorded by all kinds of mobile devices are pushed online instantly via high speed data links to be shared with people around the world. Search engines such as Bing and Google index almost everything available on the web in an attempt to give users instant access to information relevant to them. However, it is estimated that about 1 trillion queries were left unanswered despite the best efforts of search engine scientists and engineers.

The key challenge is the selection or consultation of the stored knowledge as Dr. Bush rightly pointed out years ago. Bing and Google still rely on simple user queries as the major criterion for selection to fill their search results pages. In this talk, I would like to share with you how context-aware intelligent companions might be able to address this challenge by taking advantage of ready available ambient and contextual information.

Interacting with Face
Jian Sun, Microsoft Research Asia
Face-to-face interaction is the most effective way for the communication between humans. However when users are facing computing devices, the challenge is how the computer can understand (locate, track, recognize, and even read the emotions) our faces. In this talk, I will present several applications related to the interaction between the computer and the human face: face detection, face alignment, face tracking, face recognition/tagging, and face expression capturing.

Natural User Interfaces for Activity-based UbiComp
James Landay, University of Washington
Ubiquitous Computing uses novel technology in physical activity inference and mobile device user interfaces to support people’s high-level, long-lived activities. Target activities include motivating individuals to get fit, reducing a family’s environmental footprint, and learning a second language. The everyday scenarios in which activity-based applications will be used are also better supported by Natural User Interfaces (NUIs). We will demonstrate the activity-based applications we have built as well as show how activity inference can make Natural User Interfaces work better by helping to disambiguate the inherently ambiguous input often used by NUIs.

Natural User Interactions Research: NUI Opportunities for Collaboration

Stewart Tansley, Microsoft Research
NUI is about rethinking the ways in which people will interact with computers of the future. NUI means reevaluating everything from the computer’s design to human needs and interaction models. NUI will revolutionize the way we think about computers and what they can do on our behalf. NUI represents the dawn of a new paradigm for user interaction. This talk will explore the scope of NUI, and the research challenges and opportunities ahead from the perspective of Microsoft External Research.

• Finding, Keeping, and Nurturing Talent: The Key to Success
  Session Co-Chairs: Baining Guo, Microsoft Research Asia; Weiping Li, University of Science and Technology of China
  
  The responsibility of identifying and fostering top-quality talent of the future inevitably falls to the academic and industry communities. The rapidly evolving computing world requires innovative thinking and the knowledge and determination to push beyond current-day capabilities. This talent must be identified, trained, and inspired, and universities are particularly well-suited to making such investments in research talent. The outstanding universities will be those institutions that are able to respond to the changes in future generations of students, as well as their demands. They will also need to acquire resources outside the academic environment to foster the best talent. Industry must also be concerned with attracting future research talent to ensure products and services that the world needs, as well as profitability. Collaboration between academics and industries is one way to foster future research talent.
  
  In this workshop, researchers from Microsoft Research, as well as our academic counterparts, will share best practices and discuss the challenges. 

Overview: Talent Programs at Microsoft Research Asia

Lolan Song, Microsoft Research Asia
Microsoft is committed to cultivating talent in information technology in the Asia-Pacific region by providing opportunities to broaden talents' horizons, enhance their academic knowledge, and hone professional skills. I will introduce MSRA’s talent programs, and highlight our collaboration with the top universities in the Asia-Pacific region.

Requests from Customers, and Our Efforts and Plans to the Requests
HyunWook Park, Korea Advanced Institute of Science and Technology (KAIST)
Huge amounts of information are being delivered through very fast communication channels that allow humans to easily communicate anywhere in the world. Multiple technologies have been combined and converged to unveil various ways to improve the quality of human life. These changes and pressures from our customers make us think about improving our education system as well. The demand to foster talented engineers and researchers who have various abilities from general humanity to interdisciplinary knowledge has become increasingly important in order for them to be ready to strongly contribute to their company. There may be limitations for each student or one education system to satisfy all demands from customers. We can provide multiple programs rather than one total solution to meet various demands from customers.

This talk introduces several efforts and plans of KAIST, which include small innovations in the department level and large innovations of new programs. Our plans in KAIST can suggest several trials of small and large innovation.

Recruiting, Cultivating, and Retaining Talented Academic Researchers: The Case of the Computer Science and Engineering Department at the Hong Kong University of Science and Technology
Mounir Hamdi, Hong Kong University of Science and Technology
The Department of Computer Science at the Hong Kong University of Science and Technology has recently been ranked by the The Academic Ranking Of World Universities as number 26 in the world and number 1 in Asia in computer science. It has also been ranked number 9 in the world in terms of research publications in the top 20 percent of computer science journals. We do understand that some of these rankings can be disputed. Nonetheless, they can be used as an indication of the quality of academics within a department.

In this talk, we will highlight our experience and best practices in terms of recruiting, cultivating, and retaining talented academic researchers in the Computer Science and Engineering Department at the Hong Kong University of Science and Technology as an example of the success stories as well as the challenges that face computer science departments from all over the world in this regard. Then we will draw some suggestions and strategic plans on the best way forward to achieving success in finding, keeping, and nurturing talent.

Reducing Talent Gap
Xiaoning Ling, X-Gainian Foundation
I will briefly analyze research talent requirement, and assess talent gaps between the requirement and the reality. I will then present what we have been practicing at Software School of Hunan University, China, with our intention to reduce the talent gaps. Our core approach is learning-by-doing which leads to a few new innovative curriculums and a couple of startup student projects funded by X-Gainian, an education venture capital firm. We believe our learning-by-doing approach creates an integrated platform environment where students get trained with true know-how knowledge/skills. Finally I would like to share with you my observation about a fun every-day-life training “platform," which has been effective for U.S. education.

Teaching Advanced Software Engineering
Xin Zhou, Microsoft Research Asia
Since 2007, I've been teaching an advanced software engineering course at Tsinghua University, Peking University, and Beihang University for a total of seven semesters. I tried to bring modern software engineering practice into classrooms of Chinese college. Over the course of several semesters, the result was very encouraging. I will outline the curriculum, student's feedback and my observation in this talk.

13:15–14:30

17:00–18:00