Der deutsche Education Blog

April, 2014

Microsoft Research Connections Blog

The Microsoft Research Connections blog shares stories of collaborations with computer scientists at academic and scientific institutions to advance technical innovations in computing, as well as related events, scholarships, and fellowships.

April, 2014

  • Microsoft Research Connections Blog

    FetchClimate—harnessing the cloud to find and share environmental data

    • 0 Comments

    Scientists around the world are striving tirelessly to monitor and model the environment—to understand the intricate workings of our ecosystem—so that policymakers can make informed decisions that lead to a sustainable future for “spaceship Earth.” This research involves using the thousands of available environmental datasets, on everything from agriculture and biodiversity to climate and the oceans. But finding, browsing, choosing, and downloading the right data can be ridiculously hard, even for the experts.

    What if finding environmental data were as simple as clicking on a map?

    Draw a box around the geographic area you’re interested in, select the environmental information you want, and view the data on Bing Maps within seconds
    Draw a box around the geographic area you’re interested in, select the environmental information you want, and view the data on Bing Maps within seconds

    Enter FetchClimate, a tool that makes locating environmental information as easy as searching for a hotel or coffee shop online. Just draw a box around the geographic area you’re interested in, select the environmental information you want, and view the data on Bing Maps within seconds. What used to take researchers hours, days, or even weeks can now be done very quickly—by anyone. When possible, FetchClimate calculates data uncertainty, so you know how reliable the information is, and the tool allows you to specify precisely the size of the area and the period of time for your query.

    FetchClimate runs in the cloud, on Microsoft Azure, meaning there is no physical limit on how much information can be added. You can not only look at historical climate data but also peer into the future, as we have included forecast data from the latest climate simulation experiments. For example, you can see what the predicted temperature or precipitation in your area will be in 2050.

    Visualization of year-to-year precipitation averages in southern Asia
    Visualization of year-to-year precipitation averages in southern Asia

    The Computational Ecology and Environmental Science group in Microsoft Research has spent several years developing FetchClimate, working with Moscow State University, which provided software development, and the DigiLab at the London College of Communication, which designed an interface that makes finding and understanding environmental information stress-free. So we’re excited to be releasing FetchClimate—in three different ways—for anyone to use for research, study, or just to satisfy their curiosity about our planet.

    • First, anyone can access our own FetchClimate service via a web explorer, which features a number of useful climatological layers, including temperature, precipitation, and sea depth. (If you would like to make your information and data easily available within our service, please contact us. We are interested in augmenting our current climate and environmental data with socio-economic and health information—or any other global information best viewed on a map.)
    • Second, you can access the same service via an API, from inside a program written in any of several languages, including .NET, R, and Python.
    • And third, we’re releasing a deployment package so that the more technically savvy of you can set up your own FetchClimate-powered service to make your data available to your colleagues, and to the wider world, via the same web explorer that runs over our own service. We’ll soon be open sourcing that explorer portion of the deployment, so you can customize or skin it how you like (all we ask is that you acknowledge that the service is powered by FetchClimate).

    The deployment package will be attractive to individuals, research teams, national laboratories, and international collaborations who are used to dealing with geographical data and are keen to share it with colleagues and the outside world in a more dynamic way. For example, Ireland’s Marine Institute has created the Irish Digital Ocean–SMART Marine Research Platform to stimulate collaborative research across the marine sector. As Eoin O’Grady, Information Services & Development Manager at the Marine Institute, explains, “FetchClimate greatly simplifies access to scientific data, promoting reuse. We see it as an excellent way to share Irish marine research data, part of the Irish Digital Ocean, with a broad range of users in the marine community, to support research and innovation and as input into public information services."

    In addition, we are currently sponsoring a special Climate Data Initiative that offers grants of Microsoft Azure resources to help early adopters set up their own FetchClimate-powered services. Using the deployment package, you will be able to implement your own instance of FetchClimate, including your datasets and a web front end that is customized for your own site—and we’ll provide the space on Azure! If you would like to pursue this, please submit a proposal by June 15, 2014. We will be selecting 40 awardees from among these proposals.

    We created FetchClimate as a way to turn data into actionable information, and to make that information easily available to the world. There are some exciting features that we haven’t discussed here (hint: what if you could upload a model, not just data?), and FetchClimate is just one of several exciting tools for environmental science that we are developing. All of these tools illustrate how, with a bit of imagination, we can begin to deliver research-as-a-service on Microsoft Azure. We hope these tools will help scientists, policymakers, and the public become more informed and better equipped to take care of our planet.

    Kenji Takeda, Solutions Architect and Technical Manager, Microsoft Research

    Kristin Tolle, Director of Environmental Science Infrastructure Development, Microsoft Research

    Learn more

  • Microsoft Research Connections Blog

    Cloud computing charts a path to scientific discovery

    • 0 Comments

    With the rapid rise of data-intensive scientific research—across disciplines and around the globe—scientists in Asia, as elsewhere, face massive computing needs and challenges.

    Mindful of our role in helping scientists turn big data into big discoveries, Beijing-based Microsoft Research Asia has collaborated closely with domestic and international researchers on a wide range of topics, including the environment, data modeling, biological computing, climate change, and urban computing.

    Asia embraces the cloud

    As part of these collaborative efforts, we have worked to help researchers apply Microsoft Azure, the company’s cloud-computing platform, to data-intensive scientific research. As Eric Chang, senior director of technology strategy at Microsoft Research Asia, observes, “In this era of big data, cloud computing offers scientists a platform for dealing with massive amounts of data and the growing requirements of distributed, multidisciplinary collaborations to drive new discoveries.” Here, then, are five examples of our collaborative efforts to harness the power of the cloud for scientific research.

    Understanding ecological and hydrologic processes and their interactions in large watersheds is important to a society in need of sustainable freshwater supplies. As part of a major new research program, Professor Chunmiao Zheng and Researcher Guoliang Cao of Peking University are using Microsoft Azure to support comprehensive data processing and numerical modeling of the hydrologic cycle of the Heihe River Basin, and to continue developing cloud computing as a cost-effective solution to large-scale integrated eco-hydrologic modeling.

    Numerical modeling of eco-hydrological processes in the Heihe river basin using Microsoft AzureNumerical modeling of eco-hydrological processes in the Heihe
    River basin using Microsoft Azure

    Facilitating the analysis of climate data: Sea ice is an important component of the Earth’s climate system, and coupled climate models are indispensable tools in its study. The Coupled Model Intercomparison Project (CMIP) provides a set of coordinated climate model experiments for use by climate-modeling groups. By intercomparing the resulting model outputs, CMIP can assess the mechanisms responsible for model differences, determining why similar models produce a range of responses. A total of 1.5 petabytes of model output data, including sea-ice data, was produced by more than 30 modeling groups around the world during CMIP5 (the project’s fifth phase). Unfortunately, CMIP’s current web-based data dissemination system supports only data search and download. All other necessary data processing functions must be performed by researchers in their local facilities. Professor Yuqi Bai of Tsinghua University led his group to establish an integrated research environment for archiving, searching, analyzing, and intercomparing CMIP5 data with the CMIP5 Sea Ice Data Portal. This pilot project clearly demonstrates Microsoft Azure’s value in enabling a web-based, data-intensive computing environment.

    An integrated research environment for archiving, searching, analyzing, and intercomparing climate model output data with CMIP5 Sea Ice Data Portal
    An integrated research environment for archiving, searching, analyzing, and intercomparing climate model output data with CMIP5 Sea Ice Data Portal

    Studying terrestrial ecosystems: Terrestrial ecosystems influence climate through a complex system of bio-geophysical feedback, including carbon and water exchange with the atmosphere. Honglin He, Fan Li, and Xiaoli Ren of the Chinese Academy of Sciences have been working to build a carbon-water flux data storage system for the Qinghai-Tibet Plateau ecosystem. Their system would enable model simulation and provide a platform for uncertainty analysis. The researchers based their system on Microsoft Azure’s virtually unlimited storage capacity and its data-intensive computing architecture, which can handle enormous amounts of multisource heterogeneous data.

    Improving healthcare. Professor Yan Xu of Beihang University has been conducting research on the value of using large-scale histopathology image analysis to detect colon cancer, a common and potentially deadly disease that has a huge impact on public health. While such images provide an excellent tool for detecting early-stage colon cancer, a digitized histopathological image at 40 times resolution is roughly 15,000 x 15,000 pixels. Microsoft Research Asia is applying Microsoft Azure to histopathology classification, segmentation, and clustering, a project that will help physicians improve the accuracy of their diagnoses, thereby helping to reduce costs and save lives.

    Cataloguing biodiversity: The Biodiversity Heritage Library is an international cooperative project that has scanned and openly shares more than 100,000 volumes—totaling some 43 million pages and 97 million species records. Zheping Xu of the Chinese Academy of Sciences is leading a project that will extract information from the library’s vast store of biodiversity literature, unearthing buried information about the distribution of species. The project will generate different thematic maps, enabling researchers to extract information on species distribution according to time or region, as well as to use file formats from Bing Map and other online mapping products to display multiple types of geographic information in new ways. This information can be used in efforts to further conservation efforts and wildlife management.  

    These five pioneering projects demonstrate the immense value of using Microsoft Azure in scientific research. Moreover, these early efforts strengthen our determination to bring “cloud power” to researchers from diverse disciplines.

    To learn more about Microsoft Research’s efforts to help scientific researchers accelerate their discoveries through the computational and collaborative power of Microsoft Azure, visit Microsoft Azure for Research.

    Xin Ma, Senior Research Program Manager, Microsoft Research Asia

    Learn more

  • Microsoft Research Connections Blog

    2014 Microsoft Research Awards for SEIF announced

    • 0 Comments

    Microsoft Research Awards for the Software Engineering Innovations Foundation (SEIF Awards)With the announcement of the 2014 winners of the Microsoft Research Awards for the Software Engineering Innovations Foundation (SEIF), I can’t help but reflect on the depth and breadth of research supported by Microsoft Research. Over the past few years, SEIF has not only sponsored research into core software engineering challenges but has also funded investigations into software engineering’s applications in mobile and cloud computing and natural user interface (NUI). This year was no exception to the this emphasis on depth and breadth: for 2014, SEIF invited proposals addressing core software engineering challenges as well as those that delving into the application of software engineering to the Internet of Things and large-scale cloud infrastructure. These latter two areas were explored as a partnership between Lab of Things and Global Foundation Services, respectively.

    Before I reveal the 2014 winners, let me say a bit about the diversity and quality of the proposals. We received 129 proposals, coming from nearly 30 countries and every continent (okay, except Antarctica). SEIF has truly become an internationally recognized program! Some 70 reviewers from across Microsoft conducted an internal review of the proposals, all of which were of exceedingly high quality. It was extremely difficult to choose the 12 winning projects, each of which will receive funding of US$40,000.

    Here, then, are the 2014 recipients of this Microsoft Research Award:

    • Linking Constructive and Analytical Software Verification
      Michael Butler, University of Southampton
    • Script Synthesis through Examples
      Ruzica Piskac, Yale University
    • Dependency Aware Sharing of the Internet of Things for Integrating Apps in Smart Homes
      Jack Stankovic, University of Virginia
    • Teaching Software Engineering to the Masses with TouchDevelop
      Alessandro Orso, Georgia Institute of Technology
    • Improving Datacenter Efficiency and Total Cost of Ownership with Differentiated Software Reliability Analysis and Techniques
      Onur Mutlu, Carnegie Mellon University
    • Resource Efficient Cloud Computing
      Christoforos Kozyrakis, Stanford University
    • Augmenting Testing with Performance-Aware Behavioral Models
      Yuriy Brun, University of Massachusetts Amherst
    • Kinect-based Training to Improve Balance, Mobility, and Quality of Life in Patients with Parkinson’s Disease
      Jeffrey Haddad, Purdue University
    • A Spreadsheet Model for End-User Programming with Probabilistic Data
      Scott Hudson, Carnegie Mellon University
    • Systematic Analysis of User Reviews and Usage Data for Windows Mobile Apps
      Walid Maalej, University of Hamburg
    • CoolProvision: Provisioning of Cooling Systems for Datacenters
      Thu Nguyen, Rutgers University
    • Reliable and Resilient Microgrids for Data Centers
      Reinaldo Tonkoski and Wei Sun,
      South Dakota State University

    Congratulations to the 2014 winners of this Microsoft Research Award! See the winners on the SEIF page.

    Arjmand Samuel, Senior Research Program Manager, Microsoft Research Connections

    Learn more

Page 1 of 4 (11 items) 1234