October 30, 2014 - October 31, 2014

Asia Faculty Summit 2014

Location: Beijing, China

Thursday, October 30, 2014

  • Speaker: Jeannette Wing, Corporate Vice President, Microsoft Research

    My vision for the twenty-first century: computational thinking will be a fundamental skill used by everyone in the world. To reading, writing, and arithmetic, we should add computational thinking to every child’s analytical ability. Computational thinking involves solving problems, designing systems, and understanding human behavior by drawing on the concepts that are fundamental to computer science. Thinking like a computer scientist means more than being able to program a computer. It requires the ability to abstract and thus to think at multiple levels of abstraction.

    Computational thinking has already influenced many disciplines, from the sciences to the arts. In my talk, I will give examples from Microsoft Research of how computational thinking has changed the way research is conducted in different scientific disciplines. Computational thinking has also changed what we teach in colleges and universities today. I will speak about some recent educational efforts in the United States, the United Kingdom, and China on adopting computational thinking in education, especially at the K-12 level. Computational thinking can not only inspire future generations to enter the field of computer science—it can also benefit people in all fields.

  • Speaker: Kap-Young Jeong, President, Yonsei University

    Yonsei University is leading change in Asia’s higher education by responding dynamically to educational and market trends and by serving as the vanguard in research, education, and industry ties. Interdisciplinarity is gaining momentum as a keyword because our complex, information-rich society requires flexible thinking and an ability to adapt to new input that depends on knowledge across disciplines and on an ability to process various different kinds of information. Although existing structures within universities can pose obstacles, Yonsei supports interdisciplinarity as the direction of the future and is promoting new programs in education, research, and campus infrastructures toward this end.

    Interdisciplinarity poses challenges for universities because faculty and programs are configured along lines of existing departments, fields, and disciplines. Faculty and sometimes students can be resistant to programs that are unfamiliar or that change existing patterns of funding, research, or teaching. At Yonsei University, we are implementing institutional change that promotes interdisciplinarity. For one, we are promoting liberal arts education, which by nature promotes critical thinking and emphasizes tools over content, in the process deemphasizing majors or fields of study. Our innovative Residential College program for all freshmen at our International Campus, launched in 2014, includes a holistic education program that promotes the whole individual. Yonsei’s Underwood International College, which launched 10 years ago as Asia’s premier liberal arts college, is expanding its offerings of interdisciplinary majors across all disciplines.

    On the research front, Yonsei University established the ICONS (Institute of Convergence Sciences) in 2013, which consolidated existing research centers and provided an incentive for the formation of interdisciplinary research terms drawing on faculty from diverse fields. The Institute for Convergence Technology, as well as its undergraduate, masters, and doctoral programs in the School of Integrated Technology, aims to produce a new kind of interdisciplinary-thinking young professional. On the administrative side, we are actively promoting information- and resource-sharing among our various campuses, including the main Shinchon campus, the Wonju campus, Severance Medical School, Gangnam Severance, and Songdo International campus.

  • We are used to segmenting a university by disciplines. When more academic problems require cross-disciplinary collaboration, we may have to reconsider the current system in higher education.


    • Professor Peng Gong, Professor of Tsinghua University, an ecologist
    • Professor Kap-Young Jeong, President of Yonsei University, an economist
    • Professor David S. Rosenblum, Professor of National University of Singapore, a computer scientist
    • Dr. Jeannette Wing, Corporate Vice President of Microsoft Research, a computer scientist


    • Dr. Tim Pan, University Relations Director of Microsoft Research Asia
  • Speaker: Takeo Kanade, Professor, Carnegie Mellon University

    The field of artificial intelligence has been working on computer vision—making computers able to see—since its inception. Although it is easy for humans, computerized visual recognition is much harder to achieve than originally thought. However, today computer vision technologies are expanding to many applications, some newly conceived and others with old goals but with an order of magnitude better performance. These range from applications used in daily life, such as wearable vision, to applications for medical, industrial, and scientific visual computing. Computer vision technologies are benefiting from recent advancements in microelectronics for vast processing, image sensors for capturing tiny signals, and fundamental algorithms to make sense out of visual data. Starting with some historical perspectives, the talk will discuss exciting opportunities in computer vision.

  • Speaker: Butler W. Lampson, Technical Fellow, Microsoft Research

    I have many hints that are often helpful in designing computer systems, and I also know a few principles. There are several ways to organize them:

    • Goals (what you want)—simple, timely, efficient, adaptable, dependable, yummy
    • Methods (how to get it)—approximate, increment, iterate, indirect, divide and conquer
    • Phases (when to apply them)—requirements, architecture, process, techniques

    Of course the goals are in conflict, and engineering is the art of making tradeoffs, for instance among features, speed, cost, dependability, and time to market. Some simpler oppositions are:

    • For adaptable, between evolving and fixed, monolithic and extensible, scalable and bounded
    • For dependable, between deterministic and non-deterministic, reliable and flaky, consistent and eventual
    • For incremental, between indirect and inline, dynamic and static, experiment and plan, discover and prove

    It also helps to choose the right coordinate system, just as center of mass coordinates make many dynamics problems easier. You can view the system state as a name→value map, or as an initial state and a sequence of operations that transform the state. You can view a function as code or as a table or as a sequence of partial functions. Notation, vocabulary, and syntax are other kinds of coordinates.

    In the complex process of designing systems, both principles and hints can be justified only by examples of what has worked and what has not.

  • Speaker: Christos H. Papadimitriou, Professor, UC Berkeley

    Covertly, computational ideas have influenced the Theory of Evolution from the very start. After providing a historical overview, I will discuss recent work on evolution that was inspired and informed by computational insights. Considerations about the performance of genetic algorithms led to a novel theory of the role of sex in evolution based on the concept of mixability, while the equations describing the evolution of a species can be reinterpreted as a repeated game between genes played through the multiplicative updates algorithm. Finally, a theorem on Boolean functions helps us understand better mechanisms for the emergence of novel traits.

  • Speaker: Andrew Yao, Professor, Tsinghua University

    Computer science is generally regarded as an enabling science with wide applications to other scientific fields. Increasingly, the concepts and methods of computer science are being recognized as a source of great intellectual interest, injecting fresh ideas into other scientific disciplines. Through discourses and collaborations, exciting multidisciplinary areas are blossoming. We illustrate this phenomenon from the viewpoint of theoretical computer science.

  • The field of computer science is undergoing a fundamental change. Traditional areas of computer science have been concerned with the efficiency, reliability, and scale of computer systems, attempting to make them more practical and useful—in short, to make them “work.” Traditional topics of computer science—operating systems, algorithms, databases, programming languages, and so forth—were the primary focus areas, meanwhile fully supported by theoretical computer science that developed the necessary mathematical and algorithmic foundations. Instead, modern computer science has put an increased emphasis on the processing and analysis of real-world data sets (big data) and interdisciplinary research, reaching out into areas such as economy, biology, physics, and social sciences.


    • Takeo Kanade, Professor, Carnegie Mellon University
    • Butler W. Lampson, Technical Fellow, Microsoft Research
    • Christos H. Papadimitriou, Professor, UC Berkeley
    • Andrew Yao, Professor, Tsinghua University


    • Thomas Moscibroda, Senior Researcher of Microsoft Research Asia and Chair Professor of Tsinghua University

Friday, October 31, 2014

  • Rapid global urbanization has generated great challenges, such as traffic congestion, noise, air pollution, and energy overconsumption. The field of Urban Science aims to help tackle these challenges, a task that seemed nearly impossible years ago given the complex and dynamic settings of cities. Data sensing technologies and social media have recently made it possible to accrue urban data from many sources, including human mobility, air quality, traffic patterns, and more geographical data. Cloud computing is now seen as a critical analysis tool for researchers in this field.

    What can we do with environmental data? Why are we interested in human life patterns? How does cloud technology facilitate urban computing? In this session, we bring together researchers who are targeting different urban issues to discuss their ideas and discoveries. Microsoft researchers from Redmond will also share their latest efforts on the Lab of Things, a flexible platform for experimental research that uses connected devices and cloud services to monitor and update experiments and provide easy access to collected data.

    Chair: Winnie Cui, Microsoft Research Asia


    • Yu Zheng, Microsoft Research Asia
    • Hwasoo Yeo, Korea Advanced Institute of Science and Technology
    • Victor Li, The University of Hong Kong
    • Yanmin Zhu, Shanghai Jiao Tong University
    • Takeshi Oishi, The University Of Tokyo
    • Guangzhong Sun, University of Science and Technology of China
    • Arjmand Samuel, Microsoft Research
  • Speaker: Yu Zheng, Microsoft Research Asia

    Urban computing is a process of acquisition, integration, and analysis of big and heterogeneous data generated by a diversity of sources in urban spaces to tackle the major issues that cities face, such as air pollution, energy consumption, and traffic congestion. Urban computing connects unobtrusive and ubiquitous sensing technologies, advanced data management and analytics models, and novel visualization methods to create win-win-win solutions that improve urban environment, human life quality, and city operation systems. In this talk, I will present our recent progress in urban computing, introducing the applications and technologies for integrating and deep mining heterogeneous data. Examples include fine-grained air quality inference throughout a city, city-wide estimation of gas consumption and vehicle emissions, and diagnosing urban noises with big data. The research has been published at prestigious conferences, such as KDD, and deployed in the real world. Learn more (opens in new tab).

  • Speaker: Hwasoo Yeo, Korea Advanced Institute of Science and Technology

    As the data from the roadway increases with the level of traffic congestion, prediction of travel time is emerging as the most wanted information for roadways users. It provides predicted travel times for each origin-destination pair and can suggest the best departure time to guarantee on-time arrival. In this research, we developed and implemented real-time systems to predict travel time by using the Microsoft Azure cloud. We adopted a data-driven approach that uses historic traffic sensor data to find the pattern that is most similar to the current real-time data, providing accurate prediction results with higher accuracy. The newly developed matching strategy provides a robust result with a longer horizon of prediction. We also developed a data-driven OD prediction methodology and online traffic simulator based on Cell Transmission Model, which can be used for the prediction of travel time for special cases of events with diverse scenarios. The research can be applied for the nation-scale system for travel time prediction and simulation.

  • Speaker: Victor Li, The University of Hong Kong

    Air quality has deteriorated rapidly in Hong Kong and China, with NO2 and PM2.5 levels frequently exceeding WHO safety guidelines. Although poor air quality has clear public health impacts, very few monitoring stations measure major air pollutants; there are only 13 monitoring stations in Hong Kong and 35 in Beijing. This severely limits evidence-based decision-making about air quality and leads to severe criticisms about the transparency and public relevance of the official Air Pollution Index.

    Because air pollution is highly dependent on location, and monitoring stations are costly and bulky, a citywide air quality monitoring system would be prohibitively expensive. Urban big data can be used to fill this gap. By analyzing the causality between human dynamics data (such as vehicular traffic and points of interest data) and measured air quality, we can estimate air quality at locations not covered by monitoring stations. However, processing the massive volume of data poses another challenge. To overcome this challenge, we note that most of these data are spatially and temporally correlated. Our approach is to exploit such spatio-temporal (S-T) correlation to process “part” of the data instead of “all” of the data. We detect, measure, quantify, and visualize causalities between various urban dynamics data and air quality. Causalities can be expressed in a probabilistic manner spatially and temporally. Furthermore, we exploit parallel computing by separating and allocating relatively independent data blocks to different computing resources, based on causality measures. In this way, time efficiency and scalability can be achieved. Our approach will be illustrated by using data from Shenzhen, China.

  • Speaker: Yanmin Zhu, Shanghai Jiao Tong University

    The city where we live is facing increasing challenges, such as pollution, traffic congestion, and noise. Taking the city’s pulse is to monitor the urban dynamics, and this enables people to live a better life in the city. Thanks to rapid development of the mobile Internet and various mobile devices such as smartphones, vehicles, and smart watches, mobile crowd sensing presents a new paradigm of large-scale sensing data collection. The salient features of mobile crowd sensing include the large number of data sources, large coverage, inherent node mobility, and low deployment cost. This talk discusses the basic approach of mobile crowd sensing for monitoring urban dynamics. Several mobile crowd-sensing examples will be discussed, including urban road traffic monitoring with floating vehicles, map updating with vehicular GPS traces, and urban noise monitoring with noise data from smartphones.

  • Speaker: Takeshi Oishi, The University Of Tokyo

    To help reduce CO2 emission from road traffic, we have constructed a system that prompts people to adopt eco-friendly travel behavior by modeling and showing regional traffic. The traffic flow is modeled by using street-side cameras, and CO2 emissions in regional areas are estimated by traffic simulations. The CO2 emissions are visualized with VR/MR technologies and accordingly reduced by eco-friendly behaviors. Currently, we are developing 3D city modeling techniques for more accurate traffic simulations. In this presentation, we introduce our projects in urban scenes.

  • Speaker: Guangzhong Sun, University of Science and Technology of China

    In recent years, real-world data reflecting campus life has become widely available, including users’ smart card records, mobile phone signals, GPS traces, data from cameras, and data from several management information systems. As a result, we are ready to carry out real campus computing activities that lead to a better and smarter campus. By better sensing and understanding the users on campus, we are more likely to design effective strategies and intelligent systems for improving life in campus areas. In this talk, I will present some of our research and practical works on the campus of University of Science and Technology of China.

  • Speaker: Arjmand Samuel, Microsoft Research

    An increasing number of research areas rely on collecting data from sensors and devices deployed where people live, work, and play. Researchers typically deploy sensors and devices in a few homes or workspaces, collect data, analyze the data, and make interesting inferences based on this data. Healthcare and energy management are two examples of such areas. To have confidence in the research findings, it is desirable to collect sufficient data from a large numbers of locations and in a variety of different situations and locales. However, doing so requires major investment in engineering expertise and technology infrastructure—both not readily available to the academic community.

    The Microsoft Research Lab of Things aims to provide such an infrastructure to facilitate at-scale in-situ research in a number of research areas. After its initial release in July 2013 and subsequent updates, Lab of Things is now being used in a variety of research domains. In this session, we will introduce the design philosophy behind the Lab of Things, and provide an overview of its current deployments. This session is aimed at academic researchers from diverse research areas, including healthcare, energy management, sensor design, data analysis and visualization, privacy, and system architecture design.

  • The sciences are currently undergoing a fundamental transition due to the avalanche of data that is generated by instruments, simulations, online archives, and social media. The impact of this data revolution is seen in every discipline. Cloud computing was invented to manage the big data challenges of Internet companies, but it is now seen as a critical tool for many research communities. Cloud computing makes it much easier to accrue data from many sources and to make it available for analysis by large communities.

    This session will feature academic researchers who have used cloud computing for science research projects. We highlight five projects from around the Asia region. We will also look into science research tools, new tools for machine learning, and data analysis in the cloud that we have recently made available to the community.

    Chair: Miran Lee, Microsoft Research Asia


    • Zheping Xu, Chinese Academy of Sciences
    • Tai-Quan Peng, Nanyang Technological University
    • Huayi Wu, Wuhan University
    • Hyunju Lee, Gwangju Institute of Science and Technology
    • Jun Zhu, Tsinghua University
    • Junsheng Hao, Shanghai Yungoal Info Tech Co., Ltd.
  • Speaker: Zheping Xu, Chinese Academy of Sciences

    We are experiencing the sixth mass extinction of plants and animals, and an effort is needed to protect the biodiversity of our planet. In addition to the information available from specimens and observations, there is a significant amount of information about the temporal distribution of species that can be extracted and processed from scientific literatures. However, this requires a high-performance environment to store and process the huge amounts of data, including more than 100 million records on 40 million pages. More real-time biodiversity data can be obtained from the websites of journals, news, botanical gardens, protected areas, and the communities of citizen science. This information should be integrated, analyzed, and displayed in a cloud environment that enables external users to interact with it. There are also some new techniques that should be introduced, including machine learning, natural language processing, and GIS.

  • Speaker: Tai-Quan Peng, Nanyang Technological University

    Big data is of the people, by the people, and for the people. But data could not speak for itself. The interdisciplinary collaboration between computer scientists and social scientists helps restore silent data into dynamic interaction between social topics. This project aims to examine how social topics cooperate and compete with each other to gain public attention, and to uncover what kind of factors will affect the cooperation and competition (jointly called “coopetition”) between social topics. Building on classical agenda-setting theory in communication research, the project proposes a visual analytics system that can facilitate panoramic and in-depth analysis of topic coopetition on social media. We model the complex interaction among topics as a combination of carry-over, coopetition recruitment, and coopetition distraction effects. The mathematical model provides a close functional approximation of the coopetition process by depicting how different groups of influential users (that is, topic leaders) affect coopetition. We also design EvoRiver, a time-based visualization, which allows users to explore coopetition-related interaction and to detect dynamically evolving patterns, as well as their major causes. We test our model and demonstrate the efficiency of our system based on two Twitter datasets (social topics data and business topics data).

  • Speaker: Huayi Wu, Wuhan University

    With the advancement of sensors and information technologies, a large amount of geographical information resources (GIRs)—including geodata, algorithms, application, and models—have become available on the Internet for public use. However, the heterogeneous nature and complexity of the Internet environment make it a challenge for people to discover distributed online GIRs efficiently and utilize them intuitively. GeoSquare is collaborative GeoProcessing framework designed to tackle this urgent problem by adopting Microsoft technologies. Through building a platform with integrated functions (for example, GIRs publishing and GeoProcessing orchestration), GeoSquare can help researchers and teachers share and utilize online GIRs in an efficient, harmonious way. Specifically: (1) Rich Internet Applications (RIA) technologies enrich web user interaction, (2) Azure-based cloud platform provides elastic and unlimited computing and storage resources, and facilitates global load-balancing, and (3) web service composition and scientific workflow technologies enable the online GeoProcessing orchestration and help integrate dispersed GIS functions collaboratively.

  • Speaker: Hyunju Lee, Gwangju Institute of Science and Technology

    This talk presents an extended version of DigSee, a disease gene search engine with evidence sentences. Biological events such as gene expression, regulation, phosphorylation, localization, and protein catabolism play important roles in the development of diseases. Understanding the association between diseases and genes can be enhanced with the identification of biological events involved in this association.

    Biological knowledge has been accumulated in several databases and can be accessed over the web, but there is no specialized web tool that enables a query into the relationships among diseases, genes, and biological events. For this task, we developed DigSee to search Medline abstracts for evidence sentences describing that `genes’ are involved in the development of ‘disease’ through `biological events.’ Previously, DigSee supported only cancer; now we are working to extend it to all diseases. The number of abstracts to be processed increased from 2,056,082 for cancer to 17,282,190 for all diseases, and the components that require major computational resources include crawling from PubMed, gene symbol extraction, gene normalization, and event extraction. In this talk, we will focus on extending DigSee to all diseases, including nervous system disease and cardiovascular diseases, by using Microsoft Azure. DigSee is available at gcancer.org/digsee (opens in new tab).

  • Speaker: Jun Zhu, Tsinghua University

    The growth of social media over the last decade has revolutionized the way humans interact and industries conduct business. Social media appears in many forms, including blogs, micro-blogs, forums and message boards, social networking sites, wikis, social bookmarking, tagging and news, writing communities, photo/video-sharing sites, and instant messaging. Machine learning techniques provide researchers and practitioners with the tools they need to analyze large, complex, and frequently changing data. In this talk, I will give a brief overview of social media mining, an interdisciplinary field that applies machine learning tools to social media data. I will highlight some illustrative examples done in my group, including social behavior analysis, social link prediction, and large-scale social topic graph visualization.

  • Speaker: Junsheng Hao, Shanghai Yungoal Info Tech Co., Ltd.

    Machine learning–training computer systems with historical data to predict future trends or behavior–is used in diverse applications, and more applications are being devised every day. Search engines, online recommendations, ad targeting, virtual assistants, demand forecasting, fraud detection, spam filters—machine learning enables all these modern services.

    In this presentation we will:

    1. Understand the power of Azure cloud-based predictive analytics through a short case study on language auto-detection.
    2. Demonstrate how to create a simple experiment by using Microsoft Azure Machine Learning Studio.
    3. Walk through the overall process of developing a predictive solution by using Microsoft Azure Machine Learning.
  • Computers play a major part in almost every aspect of our lives today, from the controls that let us drive our cars to our interactions with friends through social networks to health monitoring on smartphones. Sensors are now more capable, computation is cheaper and more powerful, and user interfaces are more sophisticated. Such advances have developed new human and computer interaction technologies such as the Microsoft Kinect sensor. When we look at the education space, such new technologies advance us to the next level of productivity in study and work. Meanwhile, black-boxed technology raises a new challenge, in that fewer students today are interested in computers and programming.

    This session will invite academic researchers to introduce their cutting-edge research in the field of Human Computer Interaction (HCI), such as tele-manipulation and tele-operation, second language learning, and minority language translation—providing us with some insight into the next frontier of HCI-related research. The second part of this session focuses more on educational aspects. We will introduce an interesting story of the universe of computing, new trends of programming, and a powerful authoring tool of MOOCs type of contents.

    Chair: Noboru Kuno, Microsoft Research Asia


    • Jeha Ryu, Gwangju Institute of Science and Technology
    • Sangyoun Lee, Yonsei University
    • Hiroyuki Kajimoto, University of Electro-Communications
    • Hao-Chuan Wang, National Tsing Hua University
    • Conghui Zhu, Harbin Institute of Technology
    • Darren Edge, Microsoft Research Asia
    • Kangping Liu, Microsoft Research Asia
    • Judith Bishop, Microsoft Research
  • Speaker: Jeha Ryu, Gwangju Institute of Science and Technology; Sangyoun Lee, Yonsei University

    Telepresence is humanity’s long dream. Imagine that your child is studying in a foreign country, or that you are busy working at the office but your elderly parent is in a hospital bed or is at home waiting for your visit. You want to see your child or parent, and want to hear his or her voice. As a very close family member, you may even want to touch your child’s face, or hold your parent’s hand.

    These desires encourage us to develop a truly interactive telepresence technology that involves real people over networks. The dream can be realized by using many recent technological advances, including very fast Internet that can connect anybody, anytime, anywhere in the world. This talk will present and demonstrate a tele-immersive environment that takes into account both haptic handshaking and visual eye contact. The environment has been created by developing a physical human-like avatar that has a handshaking robotic arm/hand and an accurate face pose correction algorithm with multiple cameras. The handshaking robotic arm/hand can provide various haptic sensations, such as shaking forces, hand grip pressures, and temperature while handshaking over any networks by developing a robust haptic tele-manipulation technique that can overcome time-varying natures and uncertainties in the networks and humans. In addition, to achieve realistic eye contact, an accurate face pose estimation algorithm was developed, and the texture of face was then reconstructed and rendered based on the estimated pose information. In the talk, a live demo of the system will be provided between two different locations, in China and South Korea.

  • Speaker: Hiroyuki Kajimoto, University of Electro-Communications

    Recent advances of the natural user interface facilitate the use of the whole body as a canvas for haptic interaction. This talk focuses on three major topics of the whole-body haptic interactions: how whole-body haptics enrich reality, how they affect feeling of presence and emotion, and how they induce real motion or feelings related to motion.

  • Speaker: Hao-Chuan Wang, National Tsing Hua University

    Human communication is more than speaking, and it often involves the use of multiple communication channels, both verbal and nonverbal. A scalable and reliable method to capture and analyze hand gestures as part of communication can help advance the research of interpersonal communication and the design of communication tools. In this project, we propose a way, including experimental setup and analytical techniques, to leverage the skeleton tracking capability of Kinect for studying the role of hand gestures during conversations. We demonstrate the utility of our approach through a media comparison study, showing that we can verify how different communication media (face-to-face, video, audio) affect the number of gestures that people produce and the similarity between interlocutors’ gestures. We foresee broad applications, including fast usability testing of new communication tools in the field and investigation of communication processes in situations where nonverbal behaviors may play a more salient role, such as cross-lingual communication and language tutoring.

  • Speaker: Conghui Zhu, Harbin Institute of Technology

    Asymmetry of information interactions is one of the most important determining factors in economic and cultural imbalance in different areas. In China, there are 56 ethnic groups, more than 80 languages, and about 30 different characters. Most languages have not been supported by major translation providers yet. This limits the access that Chinese minority ethnic groups have to global information and knowledge.

    Chinese minority ethnic languages present several additional problems compared with the current Chinese-English statistical machine translation (SMT) methods. First, SMT is a data-driven method, while most of these languages do not have enough training data, not to mention the high-quality parallel corpus. Second, there are almost no related basic NLP (natural language processing) toolkits supported. Last, the training progress of a practical SMT system needs huge computation that a small laboratory can’t handle. We want to build a simple but efficient translation framework for Chinese minority ethnic languages that can build a complex SMT system in an easy way.

    With help of Microsoft Research Asia, we found Microsoft Translator Hub, which meets our requirement exactly. Its minimum input is just parallel sentences, which don’t need any processing of NLP basic toolkits. By using Microsoft Translator Hub, you can build an easy translation system as long as you annotate parallel sentence pairs. Furthermore, all computation (training, testing, and deploying) is carried out on a Microsoft server, so even a mobile phone can finish the operations of building an SMT system fluently.

    To train a more efficient system, many tiny but useful things need to be done. For example, Translator Hub is designed for general language translation, and there are some conflicts between minority ethnic languages and Translator Hub that must be solved. We will need to redevelop some basic NLP toolkits of minority ethnic languages to produce high-quality parallel data. Our goal is to smoothly translate various minority ethnic languages into English, and vice versa. Uygur is the first language supported by our platform, and the performance of the Uygur-Chinese translation system based on Translator Hub is comparable to famous open source translation software. We hope that more people will join us as we develop more and better Chinese minority ethnic languages translation systems.

  • Speaker: Darren Edge, Microsoft Research Asia

    Many activities of work and life are mediated by interpersonal communication. Some forms of communication, however, are sufficiently demanding that they require significant levels of advance planning and preparation. Two examples of such auxiliary activities are preparing to deliver a presentation and learning to speak a second language. In this talk, I will show how research projects from the Microsoft Research Asia HCI research group have transformed both of these activities for the better, by helping people resume their chosen activity more easily, use their preparation time more efficiently, and learn to communicate more effectively.

  • Speaker: Kangping Liu, Microsoft Research Asia

    The availability of high-quality education is widely acknowledged as the pathway to success in modern society. In the past few years, there has been a tremendous interest in the use of MOOCs, SPOCs, flipped classrooms, and blended-learning to provide more scalable and affordable models for student learning. However, it is still hard to author interactive online lessons, and only a small fraction of faculty members create or use them. This session will introduce Office Mix, a brand new offering from Microsoft that dramatically simplifies the creation of such online lessons, including their publishing and sharing, and associated analytics. Office Mix builds on the familiarity of faculty and students with PowerPoint to create such lessons, enabling them to use the slide decks that they already have. We will also discuss use cases beyond online learning, such as sharing and communication of academic research.

  • Speaker: Judith Bishop, Microsoft Research

    When learners perceive it as fun, learning to code is more effective and sustainable. Code Hunt (opens in new tab) uses puzzles that players explore by means of clues presented as test cases. Players iteratively modify their code to match the functional behavior of secret solutions. Through a sequence of puzzles of increasing difficulty, players can learn to code or improve their coding skills. Code Hunt can also be used for contests; it was part of Microsoft Beauty of Programming in the GCR 2014. Code Hunt is used by hundreds of thousands of people, and at such a scale, the game presents challenges in keeping puzzles refreshed, analyzing statistics, and ensuring fairness. In this talk I’ll discuss the architecture of Code Hunt and present figures of usage across different kinds of puzzles and their effect on player retention and success. Finally, I’ll show encouraging figures that the game is indeed attractive for both genders.