Faculty Summit 2011

Speaker: Tony Hey, corporate vice president, Microsoft Research

The NUI revolution is well underway, and exciting new technologies are transforming how we interact with computers, what they can do for us, and who can use them. Craig Mundie, chief research and strategy officer of Microsoft, explores the technologies Microsoft is investing in and how they are changing the way we live.

Presentations:

• Putting the Cloud in the Palm of your Hand—Victor Bahl, Microsoft Research | slides
• Participatory mHealth—An Opportunity for Innovation in Healthcare, Wellness, and Research—Deborah Estrin, University of California, Los Angeles | slides

Mobile computing is the fastest growing segment of computing today. It is envisioned that in the next 10 years there will be more mobile devices sold than fixed computers. For this explosive grown to happen, a number of technical challenges need to be addressed. In this session, researchers from Microsoft Research and the academic community will explore challenges faced in the mobile computing space and outline opportunities for innovation.

Presentations:

• Innovating for Society: Realizing the Promise and Potential of Computing—Keith Marzullo, National Science Foundation
• Transforming Electricity—Rajeev Ram, Advanced Research Projects Agency – Energy | slides
• Building Secure and Resilient Systems for the Future—Howard Shrobe, Defense Advanced Research Projects Agency | slides

Representatives from federal agencies will discuss how they see information technology enabling the future in various areas relevant to their agencies’ missions—for example, science, education, health, energy, national security—and the associated federal programs and policy issues.

Presentations:

• An Overview of Microsoft Environmental Informatics Framework (EIF)—Yan Xu, Microsoft Research | slides
• Data Sharing Made Easy using OData—Alex James, Microsoft | slides
• Data Challenges in Environmental Research—John McGee, Renaissance Computing Institute | slides

Environmental Informatics aims to utilize computing technologies to help solving environmental problems. This emerging interdisciplinary paradigm accelerates the transformation from environmental data to information, to knowledge, and ultimately to social impact. The Environmental Informatics Framework (EIF) is the strategy and research solution we are developing to engage Microsoft technologies with Environmental Informatics. Focusing on data discoverability, accessibility, and consumablility, EIF enables environmental researchers to solve the interoperability problem among vast amount of heterogeneous data. EIF is built on the cutting-edge Microsoft products and innovations that support the Open Data Protocol (OData).

Presentations:

• Cryptographic Cloud Storage and Services—Kristin Lauter, Microsoft Research | slides
• Encryption as Access Control for Cloud Security—Carl Gunter, University of Illinois | slides
• The Economics of Cloud Computing: Why a Brooklyn Latte Buys a Million Unforgeable Signatures—Radu Sion, Stony Brook University | slides

New and emerging cryptographic techniques allow outsourcing of private and sensitive data to cloud operators running in big data centers. Medical data is particularly sensitive and privacy can have tremendous value in this space. Some recent projects and solutions focus on protecting patients’ privacy in several different ways, while maintaining the ability to use data flexibly for medical research and regional cost-saving efforts. Examples include protecting patients’ records through encryption and anonymizing interactions with insurance companies. Many of these same techniques are more broadly applicable to Cloud Security and Privacy issues.

Presentations:

• Making Discovery Visible: A Participatory Approach to Developing Games for Broadening Participation In Science—Kurt Squire, University of Wisconsin at Madison | slides
• Unlike Broccoli For Chocolate: Foldit, Refraction, and How Embedded Assessment is Changing the Game of Education—Seth Cooper, University of Washington | slides
• Assessment? #%&! Why Traditional Assessment Takes the Fun out of Game Design—Tracy Fullerton, University of Southern California | slides

We stand at the threshold of a new kind of education, one that leverages technology and the increasing interconnectedness of society to teach 21st-century skills; changes the way that traditional subjects are taught; provides ubiquitous platforms for assessment, content, and identity; and ultimately realizes the long-standing vision of adaptive education, games for learning, and longitudinal educational research. There is a new awareness of this potential in government, private institutions, and academia. Learn more about the various ways that we can reinvent education and how you can help.

Presentations:

• Project Roslyn: Exposing the VB and C# Compiler’s Code Analysis—Karen Ng, Microsoft | slides
• Beyond Behavior Preservation—Ralph Johnson, University of Illinois, Urbana Champaign | slides
• How Are Developers Using Refactoring Tools?—Emerson Murphy-Hill, North Carolina State University | slides

The moment we’ve been waiting for has finally arrived! The Microsoft .NET compilers are going to give access to the internals of the compilation pipeline, with handy IDE integration too! Language research on .NET will become easier than ever, permitting new kinds of refactoring and compiler/runtime research. In this talk, you’ll see how to use the new APIs to do all sorts of fancy tricks. See automatic parallelization! Thrill to cross-language cut-and-paste! Be amazed at deep semantic analyses!

This is the first of two tutorial sessions about the Kinect for Windows Software Development Kit (SDK). We will introduce the SDK and provide an overview of its contents and how developers can use them to create rich natural user interfaces.

Presentations:

• Cosmos: a Petabyte-Scale Store and Computation Platform—Pat Helland, Microsoft | slides
• Scientific Data Analysis Today: From Terabytes to Petabytes—Alex Szalay, Johns Hopkins University | slides

Today there are at least three distinct lines of research and development that have blossomed in Big Data, each with its own unique set of challenges and opportunities: (1) Big-data storage: systems that store really big (as in hundreds of Petabytes) amounts of data; (2) Big-data analytics: systems that use new information models and analytics processes to crunch really big amounts of data from multiple sources and deliver information in real or near real time; (3) Big-data hardware architectures, the networking, storage subsystems and compute runtimes that support both data storage and big-data analytics. This session starts off with talks from three leaders in the area of Big Data, specifically a talk from OSD on how BING processes petabytes of data nightly, the failure rates and issues they encounter, followed by a talk from SQL Server outlining new information processing patterns emerging in how we extract insight from data closer to the actual source, and we close with a talk from an external researcher detailing efforts to build a data analytics appliance to support data intensive research. As host I will have prepared questions to ensure the final 30 minutes of discussion is lively/productive.

Presentations:

• Semantic Knowledge for Commodity Computing—Overview—Evelyne Viegas, Microsoft Research | slides
• Ontologies for Real World—Deborah McGuinness, Rensselaer Polytechnic Institute | slides
• A Probabilistic Knowledge Base for Text Understanding—Haixun Wang, Microsoft Research | slides
• Data Challenges in the Real World—Shoshanna Budzianowski, Microsoft | slides

Semantic knowledge constitutes one of the building blocks to human understanding. For computers to become more intelligent, we need to find ways to first produce knowledge in reliable and scalable manner, and then, to use that information to derive new knowledge. This comes with several challenges, acquisition bottleneck, and inference complexity. This Faculty Summit session will discuss these challenges and investigates answers via affinity groups created over the two sessions to get semantic knowledge close to a commodity computing resource. Session 1 explores steps to acquiring semantic knowledge at scale from terabytes of data, following approaches where training data have been successfully applied to several applications such as machine translation information extraction and question answering. These resources are contrasted with richer hand crafted knowledge sources. The discussion will explore the need for a hybrid approach making usage of automation and collective intelligence to bypass the acquisition bottleneck.

Speakers:

• Eric Rudolph, Microsoft Research
• Carlos Garcia Jurado Suarez, Microsoft Research | slides

This is the second of two tutorial sessions about the Kinect for Windows SDK. In this tutorial, we will look more closely at how developers can use the APIs, sample code, and drivers that are included in the SDK to build Kinect capabilities into their applications.

Presentations:

• Part I: An Overview of Dynamic Languages: How They’re Different, Why They’re Popular, and Where They’re Being Used—Jan Vitek, Purdue University
• Part II: Recent Advances in Dynamic Languages—Jan Vitek, Purdue University

Dynamic programming environments—like Python, R, and Matlab—are the new paradigm of scientific computing, particularly for data-intensive work, due to their ease of use, instant feedback, flexible visualization capabilities, and large set of available libraries. In this session, we will describe the what, where, why, and how of this trend: what dynamic languages are and how they are different, where they are being used, why the shift is happening, how it improves productivity, the perceived versus real differences in performance, and much more.

Speakers:

• Maureen Biggers, Indiana University | slides
• Ann Quiroz Gates, University of Texas at El Paso
• Ed Lazowska, University of Washington
• Bobby Schnabel, Indiana University

Do you want to learn some simple, straightforward tactics to use to increase the gender diversity within your computing program? Do you want to learn how to share some tips and tricks with your colleagues? This session will provide practical how-to information and practices. Panelists will share results from programs that implemented research-based recruiting and retention strategic initiatives to successfully increase the number of undergraduate women the program. Participants will receive supporting resources, including workbooks, promising practice sheets, and cards with talking points. Participating faculty will be able to return to their institutions with immediately actionable activities and a framework for long-term sustainability.

Presentations:

• Text and Context: Using Context to Better Understand Searcher’s Intentions—Susan Dumais, Microsoft Research | slides
• Learning to Map Sentences to Meaning—Luke Zettlemoyer, University of Washington | slides
• Citizen Sensing—Opportunities and Challenges in Mining Social Signals and Perceptions—Amit Sheth, Wright University | slides
• Semantics and Knowledge-Enabled Experiences for Search and Discovery—M. Srikanth, Microsoft | slides

Following on from the previous session, this one will look into the use of such semantic knowledge (domain small knowledge bases) as opposed to larger scale, domain independent knowledge bases in the context of tasks such as, for instance, information mining of text, images, or videos. The discussion will contrast the amount of inference needed in each case.

• Mark Bolas, University of Southern California
• Justine Cassell, Carnegie Mellon University | slides
• Mary Czerwinski, Microsoft Research
• Daniel Wigdor, University of Toronto

Last year, we presented a slice through a representative sample of NUI research at Microsoft Research and some of our key research partnerships with academia. This year we convene a single panel to discuss the state of the art in NUI, key trends, and where we go from here as a combined research community.

The tutorial will be divided into:

• An overview of Sho: architecture, features, and capabilities
• A workshop: group programming exercises to get you up and running

In this talk, I survey a number of computer vision technologies that were developed at Microsoft Research over the last decade, including:

• Recognizing human poses and actions for range data (Kinect)
• Segmenting and recognizing organs in three-dimensional (3-D) medical images
• Stitching together dozens to hundreds of photographs for seamless navigation of panoramas, street side images, and 3-D object movies
• Recognizing manufactured objects with your cell phone for quick comparison shopping

I will also discuss some of the challenging open problems in our field.

Presentations:

• A Tour of Dafny—Rustan Leino, Microsoft Research | slides
• Experience with Using Dafny for Teaching an Introductory Course on Program Reasoning—Rajeev Joshi, NASA JPL
• Teaching Using Spec# in Europe: A Experience Report from University Teaching and Various Verification Tutorials—Rosemary Monahan, National University of Ireland, Maynooth | slides

This session will start with a tutorial on the language and program verifier Dafny. The language is object-based and including constructs for writing contracts like pre- and postconditions and loop invariants. The verifier runs in the background of the Visual Studio IDE and guides the programmer toward correct programs. One of the most promising uses of Dafny lies in teaching the concepts of program correctness to students. This session then continues with an experience report from using Spec# and Dafny at NUIM, Caltech, Nancy, and other schools and tutorials, as well as a discussion on how to share in teaching material used to teach programming using tools.

Presentations:

• Advanced Medical Imaging Research at Microsoft and its Applications on Product Groups—Khan Siddiqui, Microsoft
• Inner Eye: Toward a Computational Platform for Imaging Metadata—Steve White, Microsoft | slides
• Applications of Advanced Semantic Tagging in Clinical Settings—David Haynor, University of Washington | slides

Analysis and metadata extraction and from medical image data represent significant computational challenges, but current open source efforts in the field of medical imaging focus on sharing code rather than sharing information. A common platform enabling researchers to benchmark and integrate very different analysis techniques in a common environment, and exchange both data and analyses on the web, would greatly accelerate research in this area. In this session, the speakers will present three different aspects of how Microsoft and its partners are addressing these challenges in terms of research, development, and real-world deployment.

Presentations:

• Under the Hood of Microsoft Academic Search—Xin Zou, Microsoft Research | slides
• Eigenfactor Metrics: A Network Approach to Assessing Scholarly Journals / Eigenfactor: Ranking and Mapping Scientific Knowledge—Jevin West, University of Washington | slides
• What’s Next for Microsoft Academic Search—Adnan Mahmud, Microsoft Research | slides

Based on the original research work and prototype that was produced at Microsoft Research Asia, Microsoft Academic Search (MAS) is becoming the “go-to” search engine that researchers use to conduct in-depth searches for relevant research papers, trends, conferences, and colleagues. To help the academic community further advance their work, Microsoft Research has opened up MAS and made it a cloud-based, knowledge service platform with intelligent data inside. At this session, Dr. Wei-Ying Ma and his team members provide an overview of MAS and related research work. The session will also demonstrate how to build user-defined applications that are based on MAS and how to take advantage of the power of the community to improve the experience of academic search.

Presentations:

• Live Andes (Advanced Network for the Distribution of Endangered Species): A New Tool for Wildlife—Cristian Bonacic, Catholic University of Chile | slides
• The FAPESP-Microsoft Virtual Research Institute in São Paulo, Brazil—Carlos Henrique de Brito Cruz, FAPESP | slides
• The Brazilian Biodiversity Database and Information System: SinBIOTA—Tiago Duque Estrada, UNICAMP | slides
• LACCIR Federation: Building Research Capacity and Collaboration for ICT Innovation in Latin America and the Caribbean—Domingo Mery, Catholic University of Chile | slides

This session will provide a brief background about the organization and infrastructure of two virtual research institutes that Microsoft Research has established in Latin American and the Caribbean. One, in the State of São Paulo, Brazil, partnering with its research foundation (FAPESP), is called MSR-FAPESP Institute for ICT Research. The other is called LACCIR (Latin American and Caribbean Collaborative ICT Research Federation) and covers the entire region in partnership with top-tier research universities, the Inter-American Development Bank (IDB), and the Organization of American States (OAS). The goal of these two virtual research institutes is to foster ICT research capacity building by exploring the application of computer science to fundamental research challenges in education, healthcare, energy, and the various disciplines that are associated with environmental sciences. During this session, projects conducted by the MSR-FAPESP and LACCIR will be discussed as examples of research from this emerging region.

In designing future social experience, we explore:

• Where is social software trending?
• How does research and innovation in this fast-paced area differ from other traditional areas in computer science?
• How might we use lessons learned from social software to influence learning and education in the area of science and technology?

What we see today, the Internet and social networks connecting people via status messaging, is just a beginning. It is our hope that social software can deliver in its promise to motivate people to action, improve the quality of life and social wellbeing, and help us solve meaningful problems and create amazing things together.