-
How to write a great research paper (opens in new tab)– Simon Peyton-Jones (Microsoft Research)
How to give a great research talk (opens in new tab)– Simon Peyton-Jones (Microsoft Research)
Writing papers and giving talks are key skills for any researcher, but they aren’t easy. In this pair of presentations, I’ll describe simple guidelines that I follow for writing papers and giving talks, which I think may be useful to you too. I don’t have all the answers – far from it – and I hope that the presentation will evolve into a discussion in which you share your own insights, rather than a lecture.
New kinds of software for new kinds of science – Alexander Brändle (Microsoft Research)
Developing a precise, quantitative, predictive science of complex natural systems looms as the most active, important branch of science this century. Such a science is vital to understanding the biosphere, climate change, future food and water security, and the threat of a global pandemic, and that science could underpin a revolution in our understanding of living systems, medicine, and health. It will require new kinds of scientists to develop a new kind of natural science, and both will depend critically on radically new kinds of computational methods and tools to enable scientists to build and test dynamic, predictive models of complex natural systems, integrate them with data and experiments, and publish and share models, data, and results. This, in turn, demands new kinds of software environments to support this new kind of computational science, led by new kinds of computational scientists.
A unique collaboration between scientists and software engineers in Microsoft Research’s Computational Science Group in Cambridge is pioneering the development of these new software environments. This talk will give a short introduction into the research on tools & technologies for new kinds of science we conduct.
Presentations of past students
Hybrids of generative and discriminative models (opens in new tab)– Julia Lasserre (University of Cambridge/Max Planck Institute)
When labelled training data is plentiful, discriminative techniques are widely used since they give excellent classification results. However, hand-labelling of data can get expensive, and there is considerable interest in semi-supervised techniques based on generative models. Although the generalisation performance of generative models can often be improved by `training them discriminatively’, they can then no longer make use of unlabelled data. In an attempt to exploit the benefits of both generative and discriminative approaches, methods have been proposed which interpolate between these two extremes by taking a convex combination of the generative and discriminative objective functions. In this article, we consider that there is only one correct way to train a given model, and that a `discriminatively trained’ generative model is fundamentally a new model. From this viewpoint, generative and discriminative models correspond to specific choices for the prior over parameters, which opens the door to principled ways of interpolating between generative and discriminative extremes through alternative choices of prior. We illustrate this framework on semi- supervised learning.
Scalable display architecture (opens in new tab)– Alban Rrustemi (University of Cambridge)
Recent progress in electronic, display and sensing technologies makes possible a future with omnipresent, arbitrarily large interactive display surfaces. Nonetheless, current methods of designing display systems with multi-touch sensitivity do not scale. This talk gives an overview of the limitations of existing display systems and briefly presents the key findings of my PhD research – a platform for resolving forthcoming scalability limitations by employing a distributed architecture.
Generative face models for image understanding – Brian Amberg (University of Basel)
Humans excel at the task of image understanding. When we see a face we immediately infer if the person is male or female, attractive or not, young or old, hostile or friendly or just a boring speaker. For computers this is much harder, but some progress has been made. In this talk I’ll argue that for image understanding strong prior knowledge is needed. Humans have aquired this knowledge over the course of their phylogenesis and ontogenesis, while we are still very much trying to force-feed computers by hand. I’ll introduce you to generative face models to be used as prior knowledge, and I’ll demonstrate a range of different uses of these models which I’ve encountered during my thesis. I hope to give you an overview of the state of the art such that you know the possibilities and limitations of generative face models for image understanding.
Interactive matting (opens in new tab)– Christoph Rhemann (Vienna University of Technology)
Matting aims to accurately extract a foreground object out of a photograph or video and is an important operation in many image/video editing applications. For instance, once an object has been extracted successfully from its background by appropriate matting techniques, it may be inserted into another scene. I will give an introduction to matting and discuss recent work in this area that was developed throughout my PhD studies in co-operation with Microsoft Research. More specifically, I will discuss different models for matting and present a ground truth dataset that can be used for a quantitative comparison of matting results.
Challenges in refactoring – Mathieu Verbaere (University of Oxford/Semmle)
Refactorings are behaviour-preserving program transformations, typically for improving the structure of existing code and preparing the introduction of new functionality. A few of these refactorings have been mechanised in development environments, but many more have been proposed, and it would be desirable for programmers to script their own transformations. Correctly implementing such source-to-source transformations, however, is quite complex. In this talk, I will illustrate some common pitfalls in automating refactorings, show how to address them and how to facilitate the implementation of refactorings in general.
EU opportunities for young researchers (opens in new tab)– Carlos Morais-Pires (European Commission)
Carlos Morais-Pires, project officer in INFSO Directorate F, Emerging Technologies and Infrastructures, will present an overview of the various EU programmes supporting young researchers and students in Europe.
-
Rough guide to being an entrepreneur (opens in new tab)– Jack Lang (University of Cambridge)
At some stage you might want to exploit your ideas by starting a company, just as Bill Gates and Paul Allen did in 1975. It might even be the next Microsoft, or bought by them. I’ll give an overview of the process, explain some of the success factors investors look for, and how to go about writing a business plan and getting off the ground.
Third generation machine intelligence (opens in new tab)– Christopher Bishop (Microsoft Research)
Some of the most promising research opportunities lie at the intersections of different fields. As a small step in this direction, Chris will give a short introductory tutorial to the field of machine learning. The focus will be on the underlying concepts, illustrated with simple examples, and the mathematical content will be kept to a minimum. This talk will assume no previous knowledge of machine learning.
Parallel session
- The executable pathway to biological networks – Jasmin Fisher (Microsoft Research)
- Principles and applications of refinement types (opens in new tab)– Andy Gordon (Microsoft Research)
Tools and services for data intensive research (opens in new tab)– Roger Barga (Microsoft Research)WorldWide Telescope – A computational science innovation – Yan Xu (Microsoft Research)
The Microsoft Research WorldWide Telescope (WWT) is a computational science innovation. It sets new standard for presenting (visualizing) large data sets. WWT enables a computer to function as a virtual telescope. It brings together the imagery from the best ground- and space-based telescopes in the world. It allows users to experience, interact, and create narrated tours to feature their favorite objects in the sky for astronomical research and science education. Through the WWT Academic Program, additional software tools are delivered to enable researchers and educators to integrate WWT with their existing astronomical research and science education platforms.
Scientific computing on .NET (opens in new tab)– Jurgen Van Gael (University of Cambridge)
I strongly believe in being able to use the right tool for the right job. The .NET platform has allowed me to achieve exactly that: I’ve chosen to write most of my own code in F#, a language which gives me the flexibility to code in a functional, imperative and object-oriented style. dnAnalytics (opens in new tab), the open source numerical library I use from F#, is written in a mix of C# and C. Whenever I need to do rapid prototyping or glue together an experiment, I script it in either Python or F#. The key enabler is the .NET platform: it makes sure that all these languages understand each other. In my presentation I want to elaborate on dnAnalytics and how you can use it in your own research. dnAnalytics contains many essential tools for scientific computing: numerical linear algebra, special function evaluation, statistical tests, various distribution classes and much more. Using a short interactive demo I will highlight some of the key features of dnAnalytics.
-
How to manage your supervisor (opens in new tab) – Tennie Videler (vitae)
This session will look at how to make supervision work well for you. It will argue that to get the most out of your relationship with your supervisor you will need to be proactive and assertive. In particular the session will encourage you to think about the following issues:
- The relative rights and responsibilities of the supervisor and supervisee
- What your supervisor cares about and how you can move up in their priorities
- Strategies that you can use to manage the supervision and make them more useful
- What to do if things go wrong
Parallel session
Overview of systems and networking research at Microsoft (opens in new tab)– Tim Harris (Microsoft Research)
The research of the Cambridge Systems and Networking group at Microsoft Research Cambridge, covers the broad span of systems and networks research, ranging from improving the performance of individual computers through to designing novel distributed systems that can scale to hundreds of thousand of hosts. I will give an overview of this multi-disciplinary group that designs and builds systems, analyses them, and uses them.
New hardware enabling new user experiences (opens in new tab)– James Scott (Microsoft Research)
In this talk I will give an overview of recent research in the Microsoft Research Cambridge lab’s Sensors and Devices Group. I will cover projects such as SenseCam, a simple wearable camera which has spurred research in lifelogging and in supporting memory loss – e.g. for Alzheimer’s sufferers. I will also discuss Somniloquy, a platform enabling PCs to be put into power-saving modes more often without sacrificing functionality such as remote file transfers, by enabling them to “talk in their sleep”. I will describe Force Sensing, a way of augmenting mobile devices with pressure sensors to enable sensing of forces applied to the whole casing, e.g. twisting or bending actions, and interactions that this can enable. I will also show Second Light, a prototype surface computer which not only has an interactive surface, but which can also project images through the surface onto objects above it, and enable interaction above the surface.
Parallel session
Enabling intelligent management of the environment – Drew Purves (Microsoft Research)
In recent years all parts of society – individuals, companies, governments – have become keenly aware of the need to conserve the environmental life support systems on which we all depend. When it is fully developed, ecology will enable this conservation, by providing reliable, accurate models to predict how alternative human actions would translate into outcomes at the ecosystem level, e.g. changes in biodiversity, carbon storage and water cycling. But ecology is a young science, unaccustomed – and some would say, not yet ready – to build these predictive models. In this talk I will present some examples of the computational ecology being carried out in our group. At least one project will be drawn from each of our research foci: ecological networks; next-generation Earth System science; biodiversity and biogeography; and behavioural dynamics. What these projects share is a methodology of combing pre-existing and new concepts and models with large amounts of data, and an aim of carrying out science that is both relevant to society, and packaged in way that can be understood and used by as wide a variety of people as possible (whether fellow scientists, policy makers, or individuals).
Forza, Halo, Xbox Live: The magic of research in products – Ralf Herbrich (Microsoft Research)
In this talk, I will reveals the magic behind the Artificial Intelligence of Forza Motorsport and the machine learning algorithm Trueskill that matches players in Halo 3 and other Xbox LIVE games. I will also explain how simple geometry and advanced high-school maths mixed with tons of imagination can lead to exciting new possibilities in the realm of computer games and beyond.
Introduction to intellectual property (opens in new tab)– John Mulgrew (Microsoft)
I will present the different types of intellectual property and how those rights can be obtained. I will also discuss some of the factors Microsoft uses for determining whether potential intellectual property rights are worth protecting and when we may prefer instead to share our work openly. Finally, I will talk about some of the more common issues we encounter when collaborating with other people or using materials created outside the company.
Internships uncovered… (opens in new tab)– Peter Key (Microsoft Research)
Who is an intern at Microsoft Research? What do they do? Why should I apply for an Internship? These and other existential questions will be answered! Examples of recent Intern projects will give an insight into the variety and depth of work involved. In brief, Internships are a fantastic opportunity to learn more about research, learn about Microsoft from the inside, work with world-class researcher as mentors and colleagues, and have a great time.
-
Giving a good presentation – Ken Shaw (Benchmark Communication Techniques)
Lecture, Presentation or Conversation? We will examine: Who your audience is; What they want; Why you are addressing them; How you handle practical issues like nerves, body language, speech & voice, humour, visual aids etc.; What is success? What is plan B if everything goes wrong; How you recover.
Sustainable energy – without the hot air (opens in new tab) – David MacKay (University of Cambridge)
What do the fundamental limits of physics say about sustainable energy? The British Isles, we often hear, have `huge` renewable resources – but we need to know how this `huge` source compares with another `huge`: our huge power consumption. The public discussion of energy policy needs numbers, not adjectives. Assuming no economic constraints, assuming we cover the country with windmills and the coast with wave-machines, every roof with solar panels and every field with energy crops, could Britain get enough power from renewables to continue with our current consumption?