The demo exhibition will feature a variety of our latest technologies under development, providing an ideal environment for you to meet with the researcher’s behind these projects, hear their views on the future of computer science and try some of the technologies. You can also view some brand new technologies which haven’t been seen before.
Workloads in data centres show significant diurnal variation, with peak and trough periods. Thus, there is significant potential for saving power by making power consumption proportional to load. An effective way to do this in large data centres is in software, by turning off entire servers, and dynamically consolidating load onto the remaining servers. Storage is the main obstacle to achieving this goal. While computational state can be consolidated using techniques such as virtual machine migration, terabytes of storage per server cannot be migrated due to the high cost of data movement. This demo presents a promising new data center storage layer called Sierra, which is our answer to the above challenge. Early experiences with it show that power savings of up to 60% are possible for several of our services like Hotmail or Windows Live Messenger.
The Gathering Engine is a look into the future of web searching tools. By focussing on the perspective of the user, rather than relying on quantitative data such as clickstream or query analysis, the Socio-Digital Systems group are seeking to develop new search tools and technologies for interacting with the web. The Gathering Engine is novel in that it assumes that the user is not searching for anything in particular and is, rather, wanting to browse and gather information that seems interesting. The experience afforded is more like a dialogue between the Gathering Engine and the user, a creative exchange of possibilities and suggestion.
Portable handheld devices along with cloud computing services are predicted to have a prominent role in the near future. It is envisioned that these two previously mutually exclusive paradigms will become more symbiotic. Vector architectures cater to power-efficient execution environments with a higher performance per watt and have a strong potential in both domains. This research focuses on the applicability of highly adaptable data-level parallel architectures to multimedia and server workloads, as well as traditionally non-vectorisable workloads.
This system examines the issue of family archiving and presents a system designed to enable families to capture, manage, create and store new kinds of digital memorabilia. The system, using Microsoft Surface as its hub, shows how families can upload photos and videos quickly and easily, and scan in physical memorabilia, such as children’s artwork or a child’s first pair of shoes. The system enables families to view these media in many flexible ways and to create new, compelling kinds of digital objects, such as multimedia scrapbooks and even a digital piñata. We will further show how this system would fit into a larger ecosystem of devices in the home and link to new kinds of media displays.
TellTable builds on Microsoft Surface to provide an interactive storytelling experience, similar to how children tell stories using physical toys. Children can create any digital characters and scenarios they think of, by taking photos of real world objects, people, and environments, or by finger painting. Using multi-touch gestures to manipulate these, children can act, narrate, and record imaginative stories together, and replay the stories later to share with their friends. The system was deployed in a primary school, and received overwhelmingly positive feedback. Teachers were particularly enthusiastic about its potential as an educational tool; to help children develop communication skills and create interactive tutorials for teaching foreign languages.
Photo collages celebrate important events and themes in our lives. With AutoCollage you can just pick a folder, press a button, and in a few minutes you are presented with a unique memento to print or email to your family and friends. Developed by the Microsoft Research Cambridge Innovation Development team, AutoCollage Touch is a version enhanced for touch devices, being available pre-installed on some new Window 7 based computers. By means of natural gestures, users can fully control the application, from the selection of pictures to be considered for the collage process, to the definition of areas of interest, and inspection of the created AutoCollages.
Project Gustav is a realistic painting-system prototype that enables artists to become immersed in the digital painting experience. It achieves interactivity and realism by leveraging the computing power of modern GPUs, taking full advantage of multi-touch and tablet input technology and our novel natural media-modelling and brush-simulation algorithms. Project Gustav is a great example of how Microsoft’s research efforts are leading to exciting new technologies to support creativity.
Infer.NET is a.NET framework for writing programs which learn from data in the presence of uncertainty. The framework is unique in that allows you to combine detailed domain knowledge with the latest machine learning algorithms to generate tailored and scalable code to solve your problem. Infer.NET is used in a wide variety of research projects such as:
Identifying risk factors for asthma in health records
Inferring user intent from search click data
Identifying musical tastes and communities in Zune
The popular board game Go originated in China 4000 years ago. It only takes minutes to learn, but a lifetime to master. Although computers have surpassed human skills at Chess, creating a competitive Go program remains a research challenge. The Path of Go is powered by three technologies developed at Microsoft Research Cambridge: a state-of-the-art Go engine; the F# language; and TrueSkill™ranking and matchmaking. The entire game is written in managed code using XNA Game Studio. The game is due to be released as an Xbox 360 Live Arcade game in Summer 2010.
People can seamlessly interpret visual scenes and events, however, the search through millions of images or hundreds of hours of video becomes nearly impossible if done manually. In our work we aim to facilitate the access to visual information in very large collections of images and video. We demonstrate a trainable system that is able to automatically detect and recognise human actions in video. This, for example, is useful for sociologists who study the influence of human actions in movies, e.g. smoking, onto our society. We also demonstrate an on-line system that finds images based on their visual content in a web-scale database with ten million images.
The carbon emissions that humans have already emitted, have already affected global climate. But to head off dangerous future climate change, we need to predict future global CO2 concentrations under different ‘scenarios’ (e.g. business-as-usual, where humans carry on regardless), then predict the impact of these CO2 levels on future climate. We are developing software tools to allow for the rapid construction and testing of complex, non-linear models of this kind. Using prototypes of these tools, we recently built a new ‘balanced complexity’ model of the carbon-climate system that includes feedback resulting from land-use change, population growth and agricultural productivity.
The Peppermill project is an exploration into the design space of user interface devices that are able to source their power from the physical effort involved in interacting with them. In the prototype Peppermill device, a geared DC motor and a simple electronic circuit are used to enable interaction-powered rotary input. When turned, the circuit provides a temporary power source for an embedded device, and doubles as a sensor that provides information about the direction and rate of input. To illustrate the capabilities of the Peppermill device, we have developed a remote controlled multimedia-browsing application.
StreamInsight is a platform to enable Complex Event Processing (CEP) from data sources in a broad range of scenarios, such as manufacturing plants, web analytics and financial trading floors. Assets such as heating, cooling, ventilation machines, production and materials handling equipment can be monitored to ensure optimal and economic operation. The challenge in these scenarios is to provide close to real-time analysis and connect with backend systems in a reliable way. StreamInsight is a promising technology that can handle millions of events per second. In addition to monitoring energy consumption and production, it can manage system alarms and identify root causes of problems by using the StreamInsight Event Flow Debugger.
DNA is, in essence, the computer software of a cell. In recent years, cells have been programmed by inserting DNA code taken from other cells or synthesized from scratch. However, programming cells in a reliable manner is extremely difficult and must be done with great care. In this demo we present a software tool for designing biological circuits, simulating them on a computer, and compiling them to DNA code, which can then be inserted into living cells to program their behaviour. The tool automatically selects the DNA code that satisfies the design constraints, allowing cells to be programmed more efficiently and reliably. In future, cells could be programmed to help solve some of the biggest challenges of the 21st century, in areas of food, medicine, energy and the environment. For instance, cells could be genetically programmed to improve crop yields, fight bacterial infections, or convert carbon dioxide into energy.
DNA is also suitable for building self-contained circuits that can be inserted into cells as molecular hardware. We present a software tool for designing a range of efficient, high-performance DNA circuits, taking us one step closer to building miniaturised computers inside cells. Such circuits could be inserted into cells as "smart drugs" that monitor diseased cells and compute a response to limit the spread of viruses or cancers in the human body.
We are working on new technology to enable automatic and semi-automatic analysis of n-dimensional medical images. The results of our research will be of great help in measuring anomalies, detecting possible tumours and increasing the efficiency and accuracy of radiologists and clinicians. Ultimately, patients all over the globe will benefit from this technology.
Darwin produced six editions of The Origin of Species in his life time. A variety of changes were made as his ideas and arguments were framed within a changing scientific and cultural environment. We unlocked the Origin’s changing structure by comparing and ageing the individual sentences, paragraphs, sub-chapters and chapters through the six editions. This analysis showed where each insertion and deletion occurred, and so how the textual code that communicated the ideas evolved. Visualisation of this information demonstrates the changing form of the Origin from 1869-1872, releasing the written word from a long line of text and into an interactive, visual format.
We imagine technology heirlooms as digital objects that are precious enough to keep for longer than normal digital things, which we often tend to think of as disposable. We’re using Technology Heirlooms to explore what it means for technology to persist not just for years but for decades. They allow you, for example, to reminisce about your past as you age using digital content. Ultimately we’d hope they might become something that you pass on to your offspring. Social media sites and blogs are the new personal journal. In preserving this online content, future generations could gain insight into a relative’s life and personality through information which may have otherwise been lost.
Developed in partnership with the Microsoft Developer Division, F# is a functional programming language that extends the .NET platform and can make a significant positive impact on the lives of professional programmers, particularly those who work regularly in technical, algorithmic, parallel and data-rich fields. It enables an explorative method of programming, providing programmers with specific needs and the flexibility to translate their requirements into code quickly and easily, making the transition smoother in .NET and allowing them to spend more time thinking in their domain instead of writing code. Named F# to stand for ‘Fun’, programming with the language makes many routine programming tasks simpler and more enjoyable.
They say pictures are worth 1000 words – so it’s a great idea to use them to spice up a document or presentation. The problem is that they usually end up being self-contained rectangles in the middle of things that don’t flow with the content. Up until now the only way to isolate part of the picture was to go into a photo editing package and learn the process of removing portions of the image.
Background Removal is a new feature in Office 2010 that makes this process quick and easy for any picture. Unlike similar tools, the Office Background Removal tool doesn’t just select colour ranges or trim to a border you draw. It uses new capabilities and algorithms to achieve better results automatically with very little effort or fine tuning from the user.
We have developed a body part recognition algorithm that takes depth camera images and decides in real time which pixels belong to which parts of your body. We use machine learning to train from millions of images of people of different shapes and sizes in different poses. The resulting system can be used for touch-less, human-computer interaction and will be shipped later this year inside Project Natal for the Xbox 360.
SenseCam is a wearable digital camera that is designed to take photographs passively, without user intervention, while it is being worn. Initial clinical trials have shown that SenseCam can be a convenient and powerful memory stimulant for people with a variety of memory conditions, including Alzheimer’s disease. Originally developed at Microsoft Research Cambridge, SenseCam technology has recently been licensed to Vicon for a variety of medical applications and is now available commercially as the Vicon Revue. Members of the Microsoft Research and Vicon teams will be present to demonstrate the Vicon Revue and to describe its use in medical scenarios, as well as its potential in a number of different application areas.
In paid search (or sponsored search) the advertiser is not charged when their ad is shown, but only when a user clicks on the ad. The predicted probability that a user clicks on an ad impression (or CTR for click-through rate) is therefore a crucial quantity for optimally allocating ads to a page. Furthermore, the predicted CTR also affects the amount the advertiser is charged per click. Accurate predictions benefit both parties involved in the paid search marketplace: the user who sees more relevant ads and the advertiser who gets more clicks at a fairer price. Our team created the CTR prediction algorithm, AdPredictor, that launched with Bing in June 2009.