Microsoft Research Summer School 2008

The third Microsoft Research Summer School will be held at Microsoft Research in Cambridge on 7 July 2008 through 11 July 2008. It will include a series of talks of academic interest and posters sessions that will give invited students the opportunity to present their work to Microsoft researchers and a number of Cambridge academics.

This year the Summer School is organised in partnership with the Cambridge University Computer Laboratory. Invited students include all first year PhD students of the Computer Laboratory and all Microsoft Research 2007 PhD Scholars.

Lectures and posters sessions will be public and opened to all research staff and students from the University of Cambridge.

Agenda

Talk Abstracts and Speaker Biographies

Andrew Herbert, Managing Director, Microsoft Research

Research career - Industry vs. academia

Biography

Andrew Herbert is the managing director of Microsoft Research in Cambridge. Initially joining Microsoft Research in 2001 as an assistant director, in March 2003 he succeeded the founding director, Roger Needham.

Andrew’s research interests include networks, operating systems, programming languages, and distributed information sharing.

Before joining Microsoft Research in 2001, he was director of Advanced Technology at Citrix Systems, Inc., where he was instrumental in steering the company toward Internet thin-client technologies, initiating development of products for Web-based application deployment and for the emerging application-service-provider market.

Andrew joined Citrix in 1998 from Digitivity, Inc., which he had founded in 1996 to develop a product to enable secure deployment of Java clients for business-to-business applications. Digitivity was a spinoff from APM, Ltd., a research/consulting company Andrew had founded in 1985. APM managed ANSA, an industry-sponsored program of research and advanced development of distributed-systems technology to support applications integration in enterprisewide systems. The ANSA work included research on support for interactive multimedia services, object technology for World Wide Web applications, distributed-systems management, mobile-object systems, and security for electronic commerce. Herbert led the ANSA technical program, built the ANSA team, created the ANSA architecture, and made it known and respected in the industry.

Andrew is a fellow of Wolfson College Cambridge, England, a member of St John’s College Cambridge, and a liveryman of the City of London Worshipful Company of Information Technologists. In 1975, he graduated from the University of Leeds with a B.Sc. in computational science, and in 1978, he received a Ph.D. from the University of Cambridge in computer science.

Simon Peyton Jones, Principal Researcher, Microsoft Research

How to write a great research paper [PPT]
How to give a great research talk [PPT]

Abstract

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.

Biography

Simon Peyton Jones, MA, MBCS, CEng, graduated from Trinity College Cambridge in 1980. After two years in industry, he spent seven years as a lecturer at University College London, and nine years as a professor at Glasgow University, before moving to Microsoft Research in 1998. His main research interest is in functional programming languages, their implementation, and their application. He has led a succession of research projects focused around the design and implementation of production-quality functional-language systems for both uniprocessors and parallel machines. He was a key contributor to the design of the now-standard functional language Haskell, and is the lead designer of the widely-used Glasgow Haskell Compiler (GHC). He has written two textbooks about the implementation of functional languages. More generally, he is interested in language design, rich type systems, software component architectures, compiler technology, code generation, runtime systems, virtual machines, garbage collection, and so on. He is particularly motivated by direct use of principled theory to practical language design and implementation – that’s one reason he loves functional programming so much. He is also keen to apply ideas from advanced programming languages to mainstream settings.

Alessandro Duminuco, 2005 Microsoft PhD Scholar, Institut Eurécom

Towards a peer-to-peer file backup system [PPT]

Abstract

Peer-to-Peer systems have received a lot of attention in recent years. The key property of Peer-to-Peer systems is self-scaling, i.e. as more peers become part of the system not only the service demand increases but also the service capacity. The research community has shown an increasing interest in the use of P2P systems for file storage. This application can be very attractive for two main reasons: (i) centralized solutions are expensive (ii) common PCs are equipped with high-capacity local disks, often underutilized. The design of a Peer-to-Peer storage system is not a trivial task and presents a considerable number of challenging problems. I will illustrate briefly some of the issues involved, explaining some of the solutions I proposed so far during my Ph.D. thesis.

Fabien Corblin, 2005 Microsoft PhD Scholar, Université Joseph Fourier

Generalizing the discrete analyses of genetic networks using constraints

Abstract

Our work generalizes the current existing discrete approaches for analyzing the properties of genetic networks as proposed by Thomas using concepts that are available in constraint programming (CP). Thomas’ networks can be formalized, generalized and implemented using constraints. At the limit the proposed approach combines both aspects of simulation and reverse engineering. The ultimate goal of this work is to allow biologists to explore the combined effects of various types of hypotheses such as the assumed gene interactions and the expected dynamic behavior. We have developed a constraint program that utilizes the assumed hypotheses expressed as data. When the data is interpreted by a CP processor the proposed system is capable of responding to multiple queries that encompass simulation, reverse engineering and hybrid combination of the two. After a presentation of our approach, I will show you the constraint program in action with the following functionalities: combination of simulation and reverse-engineering, addition of hypotheses about the combination of interactions over the genes of the network, creation of mutants, relaxation of constraints in case of incoherence (the data having contradictory effects), and consequently suggestion of refinement of the model.

Georg Weißenbacher, 2005 Microsoft PhD Scholar,  ETH Zurich

Why breaking other kids’ toys (in time) is an achievement in computer science [PPT]

Abstract

Automated software verification is a superb challenge. Unfortunately, in the many years of research in this field, few actual products (if any) have been proven correct beyond doubt. We can, however, look back to a long history of spectacular successes in software falsification (think Ariane 5). The problem is that bugs are usually buried deep in the software and traditional testing techniques just don’t dig deep enough. I will talk about an approach that accelerates the digging, allowing us to dig deeper for bugs and ""break the toy"" before somebody else breaks it unintentionally. I’ll also explain how finding deep bugs in abstract software models can help verifying the actual software.

Kai Kohlhoff, 2005 Microsoft PhD Scholar, University of Cambridge

Computational prediction of three-dimensional protein structure from NMR chemical shifts [PPT]

Abstract

Proteins are involved in virtually every biochemical reaction in a living cell. A protein’s function is strongly dependent on its three-dimensional structure and there is a large interest in the pharmaceutical industry to understand the native structure of proteins in order to perform targeted drug design. Existing experimental techniques offer high spatial resolution, but tend to be costly and time-consuming. A novel method, CamShift-MD, will be presented that is based on relatively easy to perform chemical shift measurements from nuclear magnetic resonance spectroscopy. This method predicts protein structure by introducing restraints into molecular dynamics computer simulations. CamShift-MD offers a new way of automatically generating protein structures based on a combination of experiment and theory. It has the potential to solve structures for which traditional methods are not suitable, such as highly dynamical non-native states. Moreover, the method can be used to determine protein ensembles. Unlike a single, rigid structure, such ensembles reveal information about the distribution of conformations and dynamics of molecules in solution, providing us with a better picture of the behaviour of proteins in the cellular environment.

Ken Shaw, Benchmark Communication Techniques

Giving a good presentation

Abstract

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.

Biography

Ken Shaw has taught communication skills to MBA students at Judge Business School for 17 years. He has also taught at Said Business School, London Business School, Cass Business School, Henley Management College, ESMT in Berlin, Bled School of Management plus Leicester, Nottingham and Bristol universities. He has delivered training for commercial clients in France, Germany, Switzerland, The Bahamas, Holland, Norway, America, The Caribbean, Australia and Singapore.

Peter Robinson, Professor, University of Cambridge

Mind-reading machines

Abstract

Facial displays are an important channel for the expression of emotions, and are often thought of as projections of a person’s mental state. Computer systems generally ignore this information. Mind-reading interfaces infer users’ mental states from facial expressions, giving them a degree of emotional intelligence. We use video processing to track two dozen features on the user’s face. These are then interpreted using statistical techniques through a hierarchy of analyses as basic actions, head and facial gestures, and finally groups of mental states. The talk will describe an implementation of facial affect inference, together with an evaluation and some preliminary results of using the system to monitor car drivers.

Biography

Peter Robinson is Professor of Computer Technology and Deputy Head of Department at the University of Cambridge Computer Laboratory in England, where he leads the Rainbow Group working on computer graphics, interaction and electronic CAD. He is also a Fellow, Praelector and Director of Studies in Computer Science at Gonville & Caius College where he previously studied for a first degree in Mathematics and a PhD in Computer Science under Neil Wiseman.

Professor Robinson’s research concerns problems at the boundary between people and computers. This involves investigating new technologies to enhance communication between computers and their users, and new applications to exploit these technologies. The main focus for this is human-computer interaction, where he has been leading work for some years on the use of video and paper as part of the user interface. The idea is to develop augmented environments in which everyday objects acquire computational properties through user interfaces based on video projection and digital cameras. Recent work has included desk-size projected displays and inference of users’ mental states from video images of their faces.

Jack Lang, Lecturer and Entrepreneur, University of Cambridge

Rough guide to being an entrepreneur [PPT]

Abstract

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.

Biography

Jack Lang is a serial entrepreneur and business angel with high-tech and internet companies based in Cambridge, where he is Entrepreneur in Residence, Centre for Entrepreneurial Learning, at the Judge Business School, University of Cambridge, an affiliated Lecturer and member of the Faculty Board at the Computer Lab and a by-fellow of Emmanuel College. His latest venture is as co-Founder of Artimi, which is making the next generation of ultra wide band wireless chips. Previously he was founder of Netchannel Ltd, an early Interactive TV company which was acquired by ntl: where he became Chief Technologist. Before that founder of Electronic Share Information Ltd, one of the first online brokerages, acquired by E*Trade Inc. Prior to that he started a consultancy (now SAIC UK Ltd) that was involved in the early days of the "Cambridge Phenomenon", and was a proper academic at the Computer Lab. He is author of "The High Tech Entrepreneurs Handbook" (FT.Com/Prentice Hall 2001). He has other interests in molecular gastronomy and fireworks.

Tony Hoare, Principal Research, Microsoft Research

Grand challenges in computer science [PPT]

Abstract

One way to do great research is to work with a great team, collaborating over an extended timescale, to find answers to fundamental questions that lie at the very basis of your branch of science. I will call such a project a Grand Challenge.

Grand Challenges have long been common in Astronomy (the Hubble telescope) and in Physics (the large hadron collider), and more recently in Genetics (the enumeration of the human genome).  Here are six Grand Challenges that have recently been singled out by UK computer scientists for the attention of the research community:

• ‘in vivo = in silico’ – multi-level simulation of the behaviour of biological organisms;
• ‘memories for life’ – rapid recall of a lifetime’s memories;
• ‘architecture of the mind and brain’ – simulation of the relation between them;
• ‘non-classical computation’ – a search for a theory that covers the wide range of computing phenomena in nature;
• ‘ubiquitous computing’ – principles of engineering of the swarms of computers that surround us;
• ‘dependable systems evolution’ – programs that are proved correct by the computer before running them.

Can you think of any more? I will talk in more detail about the last of these, which has been my own life-long research interest.

Biography

Tony’s computing interests were stimulated by his first (and only) degree in the humanities (1956): he studied Latin and ancient Greek, followed by philosophy, with particular interest in mathematical philosophy and logic.  He learnt Russian during National Service in the Royal Navy. He spent a postgraduate year studying statistics at Oxford and another at Moscow State University, where he discovered the sorting algorithm Quicksort.  In 1960, he joined the British Computer industry as a programmer, eventually rising to the rank of Chief Engineer.

His Academic career started in 1968 with appointment as professor at the Queen’s University, Belfast.  He chose his long-term research area as proof of the correctness of programs.  In the thick of the troubles, he built up a strong computing department, and moved in 1978 to do the same at Oxford.  Following the example of Theoretical Physics, his interests broadened to the pursuit of Unifying Theories of Programming.

On reaching retiring age, he accepted an offer of employment at Microsoft Research in Cambridge, where he has seen a strong surge of interest in automation of computer proofs of program correctness.  He continues to pursue this interest, while exhorting academic researchers in long-term pursuit of even more idealistic scientific goals.

Richard Black, Research Software Development Engineer, Microsoft Research

How does the Internet work? [PPT]

Abstract

Could you explain how the Internet works in simple terms to someone else? Are there any gaps in your understanding, or questions you have? In this talk I’ll be giving an introductory description to how it all works.

Biography

Richard is currently a Research Software Development Engineer at Microsoft Research Cambridge, where he is part of the Systems and Networking group. His research interests include performance analysis of distributed systems, operating systems and networking. Recent projects include: Constellation, and Network Inference. The Network Inference project gave rise to the Network Map feature of Windows Vista. Underlying the Network Map feature is the LLTD protocol which is licensed by many companies as part of the Windows Rally program.

Richard obtained his B.A. in Computer Science from the University of Cambridge in 1990 and his Ph.D. addressing issues in operating systems and networking interaction in 1995. After a further three years at the University of Cambridge Computer Laboratory as a research associate and research fellow he moved in 1997 to a lectureship at the University of Glasgow Department of Computing Science. He returned to Cambridge in January 2000, to join the Microsoft Research laboratory, initially as a Researcher. He changed role to Research Software Development Engineer in 2006.

Wouter Spek, Director, EuroBioFund, European Science Foundation

How to apply for funding? In search of the holy grail [PPT]

Abstract

Wouldn’t it be nice to have unlimited sources of funding for your research? Wouldn’t it be nice to have the freedom to operate? Wouldn’t is be nice not to be bothered with constant applying and competing for financial sources? These are some of the topics we will discuss, we will talk about unmet needs and I will give you some pointers how to address the issue of finance and finance acquisition.

Biography

Dr Wouter Spek is currently the Director for EuroBioFund at the European Science Foundation, which he joined early in 2006. The aim of EuroBioFund, also supported by the European Commission, is to promote the development of strategic European Research Programmes in the area of life sciences. Dr. Spek received his Ph.D. in biochemistry from Leiden University (1988), with part of the research undertaken at the University of California, San Diego. From there he began his professional career at Senter (1988-1996), the agency of the Ministry of Economic Affairs responsible for all science and technology subsidy programmes, as project advisor/deputy head of the biotechnology division. Dr. Spek subsequently moved to The Netherlands Organisation for Scientific Research, where he was co-ordinator of the Life Sciences Research Council Earth & Life Sciences (ALW-NWO) (1996-2002). In 2002 he joined the Netherlands Genomic Initiative as Manager for Innovation and International Affairs, where he was involved in bringing together relevant partners for projects, including industrial partners, government agencies, and start-up companies.

Tristram Hooley, Senior Network Manager, vitae

Take control or how to manage your supervisor [PPT]

Abstract

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.

Biography

Tristram Hooley is Senior Network Manager for the UK GRAD Programme where he has responsibility for liaising between the programme and universities. He wrote a PhD on dystopian fiction before going on to work in various capacities including web design, learning technology, researcher, and skills trainer. He is particularly interested in the career development of researchers, online research methods, social capital and the pedagogy of the research degree.

Dushyanth Narayanan, Researcher, Microsoft Research

Write off-loading: Practical power management for enterprise storage [PPT]

Abstract

In enterprise data centres power usage is a problem impacting server density and the total cost of ownership. Storage uses a significant fraction of the power budget and there are no widely deployed power-saving solutions for enterprise storage systems. The traditional view is that enterprise workloads make spinning disks down ineffective because idle periods are too short. We  analyzed block-level traces from 36 volumes in an enterprise data centre for one week and concluded that significant idle periods exist, and that they can be further increased by modifying the read/write patterns using write off-loading. Write off-loading allows write requests on spun-down disks to be temporarily redirected to persistent storage elsewhere in the data centre. The key challenge is doing this transparently and efficiently at the block level, without sacrificing consistency or failure resilience. We describe our write off-loading design and implementation that achieves these goals. We evaluate it by replaying portions of our traces on a rack-based testbed. Results show that just spinning disks down when idle saves 28–36 % of energy, and write off-loading further increases the savings to 45–60 %.

Biography

Dushyanth Narayanan is a researcher in the Systems and Networking group at Microsoft Research, Cambridge. His most recent research interest is in reducing the energy consumption and improving the performance of enterprise storage.

Steve Hodges, Principal Hardware Engineer, Microsoft Research

Surface computing: the post-PC experience [PDF]

Abstract

The long-established ‘desktop’ computing metaphor continues to drive a lot of our interactions with computers, But at the same time, new interaction paradigms are being developed and we believe that these will increasingly complement the traditional computing experience. To this end, at Microsoft Research we are exploring a number of new interaction technologies along with the novel form factors, user experiences and usage scenarios they enable. This talk will present some of our work on ‘surface computing’, a new approach to interaction which doesn’t required a keyboard or mouse for input, but which is inherently multi-user and which allows direct manipulation of digital content with hands, fingers and via tangible objects.

Biography

Steve has a background in a broad range of computer-related technologies. His first degree is in Computer Science with Electronic Engineering, from University College London, and he received his PhD from Cambridge University Engineering Department in the area of Robotics and Computer Vision. Prior to joining Microsoft Research, he was the Technical Director of the Cambridge Auto-ID Lab, where he was involved with many activities, including the global research programme, the US Field Trial, the hardware-oriented action groups and privacy and IP policy development. Due to its success, the work of the Auto-ID Lab has been handed over to GS1 (formerly UCC.EAN, the barcode custodians). Steve has co-authored and presented a number of Auto-ID Centre white papers, and he has also given presentations to a wide range of audiences on the Auto-ID and EPC system. He was a founding director of the easyEPC RFID training and consultancy company. He worked as a Research Engineer for a number of years at the Olivetti and Oracle Research Lab, which became AT&T Laboratories Cambridge before eventually shutting down in 2002. This lab was internationally recognised as a centre of excellence across a broad range of advanced research into communications, multimedia and mobile technologies.

Klaus-Peter Zauner, Lecturer, University of Southampton

Enzymatic computing [PDF]

Biography

Klaus-Peter Zauner is a Lecturer in the Science and Engineering of Natural Systems Group of the School of Electronics and Computer Science at the University of Southampton. He was born in Stuttgart, soldered together a Sinclair ZX81 as his first computer and went on to study Biochemistry at the University of Tuebingen. Intrigued by Nature’s molecular scale information processing mechanisms he left Tuebingen in 1992 for Detroit to join Michael Conrad’s Biocomputing Group. Under Michael Conrad’s mentorship he worked on conformational computing and enzymatic computing. Klaus-Peter received his Ph.D. in computer science from Wayne State University, Detroit in 2001. He started his academic career as a Visiting Assistant Professor at Wayne State University, then returned to Europe in 2002 to work with Peter Dittrich in the Bio Systems Analysis Group at the University of Jena, before taking up his current position in 2003.  He served on the Governing Board of the International Society of Molecular Electronics and BioComputing and is an Editorial Board Member of the International Journal of Unconventional Computing. He is a Microsoft Research European Fellow (2005) and recipient of a Leverhulme Research Leadership award (2007).

Posters Session 1

1. Adaptive combinatorial search
   Alejandro Arbelaez (INRIA)
   

   Show/hide abstract

   The main goal, is to set automatic tuning methods for CSP algorithms, allowing e-scientists who have little knowledge of the search technique itself to nevertheless solve their optimization problem without the need for some optimization engineer. We will address both off-line and on-line tuning issues, at problem level as well as instance level. These techniques will be tested on several computational biological problems.
2. The cycle of modeling, modeling of the cycle
   Alida Palmisano (CoSBi)
   

   Show/hide abstract

   Computer Science has been used to support Biology in the storage and analysis of huge amounts of data: this is a passive usage of CS. But Computer Science can also actively help Biology in creating new knowledge by modifying the way in which the process of modeling is carried out. Starting from wet experiments, we try to standardize and automatize each step of the modeling process. We developed new conceptual and computational tools that foster a new philosophy in life science investigation. Our tools can help biologists to carry out in-silico experiments and, analyzing the results, they can suggest new experiments and/or solutions. We applied this different modeling approach to a well characterized biological model of the budding yeast cell-cycle, obtaining results that have an experimental evidence and that are not observable with the classical mathematical approach.
3. Adaptive evolutionary computation
   Álvaro Fialho (INRIA)
   

   Show/hide abstract

   Among the search algorithms applied to e-Science problems, there are the Evolutionary Algorithms (EA), that have already proved and have been constantly proving their great applicability in different areas of combinatorial optimization. But to achieve good performance, for each new problem/field, there is the need of fine tuning the EA parameters, which is a tedious and time-consuming task that requires knowledge about the domain and/or about the algorithm. Because of this, to improve the applicability of EAs to e-Sciences, this project aims to develop approaches which are able to automatically set-up their parameters, according to the characteristics of the problem or class of problems. Preliminary results are presented in the scope of the first direction taken, the Adaptive Operator Selection.
4. Adaptive software lock elision
   Amitabha Roy (University of Cambridge)
   

   Show/hide abstract

   Most software transactional memory implementations execute code using fine-grained optimistic concurrency control. This does not perform well with low contention data structures where fine grained conflict detection means manipulating metadata for every object touched and optimistic concurrency control imposes the overhead of making thread private shadow copies. Also, a purely optimistic approach does not coexist naturally with legacy code that is either already concurrent using locks or does IO operations that cannot be revoked. My proposed solution is that concurrent software can dynamically adapt itself to available parallelism using only a software runtime and without recourse to special hardware or design effort on the part of the programmer. I intend to design, build and demonstrate a practical runtime that can be used to achieve this through software lock elision. The key idea will be to separate the notions of correctness and performance in concurrent programs. The responsibility for corectness will continue to lie with the programmer who needs to specify concurrent programs using the familiar and easy abstraction of the lock. The runtime chooses the appropriate level of concurrency by eliding locks and either executing critical sections speculatively for better throughput under high contention or grouping locks into a coarser lock for lower latency under low contention. Other than better performance, such runtime supported software lock elision can be used to safely run code with potential deadlocks as well as helping to mitigate the problem of lock convoying, that has long been a criticism of lock based concurrency control. I also show how my proposed mechanism, can preserve important properties of individual lock implementations such as fairness and hence allows existing programs using FCFS locking schemes to retain fairness across threads even when used with lock elision.
5. Exact JPEG recompression and forensics using interval arithmetic
   Andrew Lewis (University of Cambridge)
6. Highly sensitive de novo identification of peptides from tandem mass spectra by linear optimization and kinetic modeling
   Axel Rack (Freie Universität Berlin)
   

   Show/hide abstract

   Protein identification (ID) using tandem mass spectrometry (MS) is a key method to study the protein content in biological samples (e.g. blood). In a clinical context, ID can put disease-related differences between the proteomes of e.g. healthy and unhealthy individuals into perspective of biological functions (e.g. metabolic pathways). Traditionally, there are two approaches for MS-based protein ID: sequence database-dependent screening and de novo prediction. The first approach is fast and relatively robust, but fails to identify novel/unknown proteins that are not stored in the databases. In contrast, the latter allows identifying virtually any protein. However, available de novo algorithms show high error rates as they usually rely on aggressively filtered data, idealized scoring models (due to limited understanding of peptide fragmentation), or simply neglect important biological information like protein alterations. Thus, current methods are unsatisfactory and show limited practical use. This work deals with the development of a new de novo method that is highly sensitive, more robust, and incorporates relevant biological and physical features. Furthermore, it must bear the computational efficiency to keep track with high-throughput data generation pipelines, which often produce noisy and incomplete data – two problems that also must be dealt with appropriately. Our approach builds upon the classical idea of spectrum graphs to efficiently perform search space pruning. Essentially, this approach translates the de novo ID problem into one of finding the highest scoring path in a directed acyclic spectrum graph. We hypothesize that satisfying solutions to the de novo ID problem will be available if a scoring scheme is used that builds upon most recent findings of peptide fragmentation rules and kinetics. This will generate large problem instances and involve scoring of features dependent on the complete peptide. Consequently, the classical spectrum graph optimization approach using dynamic programming is rendered useless, as it requires independent per-edge scores. We will address these large optimization problems by formulation of a mixed-integer linear program (MILP). However, under the assumption of a perfect scoring scheme, the influence of heuristics in the process of solving will become crucial because the solutions might deviate from the global optimum (i.e. the correct solution). We expect that even under improved - though imperfect – scoring, significantly better de novo predictions can be obtained if the MILP solver is specifically adjusted to the structure of the problem. This new de novo method will extend our highly sensitive MS processing pipeline that generates complex spectra with numerous peaks to provide superior predictions when compared to current state-of-the-art approaches.
7. Illuminating circadian rhythms with Bayesian inference
   Ben Calderhead (University of Glasgow)
8. Code reprogramming and dissemination in mobile WSN
   Bence Pasztor (University of Cambridge)
   

   Show/hide abstract

   Wireless Sensor Networks are spreading, and in many cases, these systems are deployed in remote areas for environmental/wildlife monitoring purposes, and involve tens or even hundreds of sensors. They are expected to work without much human intervention, and for as long as possible. So far, the majority of these networks are not reusable or adaptive, in the sense that once the sensors are deployed, they cannot easily be reprogrammed unless someone collects them and connects them to a computer. This task is even more difficult (if not impossible) if the devices are attached to mobile entities or animals as these cannot easily be recollected. The current (few) solutions of the distribution of new code on a mobile sensor networks involved epidemic-like spreading of the code over the network. The poster presents a group-based code-dissemination protocol, which takes advantage of the dynamically built knowledge about the social relationships between the nodes to efficiently distribute the code; and a proposed system to selectively update specific subsets of the network.
9. Quality-oriented handover scheme for adaptive multimedia streaming in heterogeneous wireless network environment
   Bogdan Ciubotaru (Dublin College University)
   

   Show/hide abstract

   Abstract—Inter-network mobility is achieved by allowing a mobile node to change its point of attachment to the network while preserving connectivity to its corresponding nodes. It is desired to achieve a high quality of service during network handover which requires the minimization of packet loss and delay which otherwise would affect negatively service quality. Streaming time sensitive and bandwidth hungry multimedia to mobile devices over wireless networks increases the challenge of performing seamless handover. Most handover solutions proposed in the literature rely on tunneling the data stream from the old access point to a new one or on switching the connection from one access point to another. These solutions involve a certain amount of quality degradation due to increasing loss and delay and suffer in terms of scalability and resilience to different mobile node speeds. Multimedia Mobility Management System (M3S) is mobility management framework which aims at maximizing user perceived quality by using multiple simultaneous connections to deliver high quality multimedia content to mobile users. M3S uses Smooth Adaptive Soft-Handover Algorithm (SASHA) as the core handover management solution which increases the quality of the multimedia delivery process when performing handover in heterogeneous wireless environment by gracefully transferring the load from one connection to the other.
10. Probabilistic interpretation of figures of speech
    Ekaterina Shutova (University of Cambridge)

    Show/hide abstract

    It is common that the author of a text in a natural language does not explicitly state all the information necessary for understanding of the text. The reader needs to address his own background knowledge or make assumptions about the world in order to interpret and reason about the text and thus approach its meaning. The key idea underlying my project is that this necessary knowledge about the world as well as the relevant linguistic information is to a large extent contained in text corpora and, therefore, can be automatically induced from them. My project concentrates on a subtask of interpretation of figures of speech using a probabilistic model. Consider an example of a metonymic phrase "enjoy a book". It is obvious to the reader that its meaning extends to "enjoy reading a book" or "enjoy writing a book" depending on the context. This, however, can not be automatically deduced in a similar way by a computer, as it is not aware of the fact that "enjoy" takes an event type argument. Nevertheless, the correct meanings can be induced from a large amount of textual data by looking at the verbs that complement "enjoy" and the verbs that take "book" as an object. The phenomenon of metaphor, although linguistically different, is addressed in a similar fashion. In metaphoric expressions seemingly unrelated features of one object are associated with another object, e.g. "I invested a lot of time in this work", "my heart was dancing". Besides making our thoughts more vivid and filling our communication with richer imagery, metaphors also play an important structural role in our cognition. Therefore, another interesting aspect of this study would be to look at the way metaphors organize our conceptual system, in terms of which we think and act.

Posters Session 2

1. Video abstraction and stylisation
   Christian Richardt (University of Cambridge)
   

   Show/hide abstract

   Video abstraction and stylisation covers a wide variety of artistic styles for simplification and expressive rendering of video sequences.
2. Time-aware routing in wireless sensor networks
   Daniele Borsaro (University of Cambridge)
   

   Show/hide abstract

   The main scenario is the one of a naturalistic park in which a network of wireless sensors is deployed with a main duty of environmental monitoring. Sensors are in part fixed and in part mobile: they collect specific data, partially elaborate them, send them to sinks for centralized and/or constraint- less elaboration, collaborate to disseminate information to interested nodes which can be data collectors or actuators. Such a network is actually a wireless sensor and actor network which topology implies a non trivial routing: nodes can be mobile or fixed; sensor or actuator or both; some can be strictly constrained on buffer size, power supply, radio range and computational power while other not. This high heterogeneity traduces in a complex network. In this first year, I am currently focusing on the fixed node subnetwork. These nodes are spread on a broad area, only few of them can be wired to an external network, many are likely to be physically accessed by staff very few times over a long period. This means that memory and power management are main concerns during nodes workout. One of the main way to save battery power is to switch the wireless interface off when the communication is not needed as, for these devices, the network communication is by orders of magnitude more expensive than the other local activities. Considering that each node has this kind of duty-cycle, the fixed nodes subnetwork results to be intermittently connected. We assume that a network manager assigns a specific duty-cycle to sensor nodes, which is disseminated in a specific startup phase. Then sensors start to exchange their duty cycle and construct their future communication behavior on the duty cycling information propagated by the neighbours. We construct a decentralized time-aware delay tolerant routing protocol which aims to provide data delivery to specific points  and minimize end-to-end cost, considering that no contemporary end-to-end path could exist between a source-destination node pair. I am currently verifying the algorithm correctness and running it by means of simulations to identify its basic properties.
3. Towards optical PCI
   David Miller (University of Cambridge)
   

   Show/hide abstract

   The local interconnect is perhaps not the first thing that comes to mind when considering the performance of a computer system, but it’s becoming apparent that electronic interconnects are going to run out of steam in the middle future. A lot of work has been done on optical cross connects as applied to network packet routing - especially with respect to power consumption - but as far as we know, there is little work in the area on local interconnects.  Where optical switching has not (yet) found much success in the internetwork switching world, there are a number of distinguishing features of local interconnects that perhaps make an optical cross connect very well suited to future performance requirements, perhaps saving power at the same time. Still, optical interconnects are no silver bullet.  There are some significant challenges (not least of which, the lack of practical optical buffering) that will require different techniques from those used in conventional interconnects.  My research is concerned with how optical switching could be used to make a cost effective local interconnect of sufficient performance to meet demands of future computer systems.
4. Spread-spectrum computation
   Derek Murray (University of Cambridge)
   

   Show/hide abstract

   Fault tolerance is important in distributed computation: processing nodes may crash, become disconnected or experience long network delays. However, existing approaches, based on rescheduling failed tasks or scheduling multiple instances of each task, are inefficient. Furthermore, they are only suited to the small class of embarrassingly-parallel algorithms. In this poster, we introduce spread-spectrum computation: a novel approach to fault tolerant distributed computation, based on the redundant encoding of algorithm inputs. The redundancy allows a certain fraction of nodes to fail, and yet the correct result can still be computed. We introduce two key concepts in this work: computation dispersal algorithms (CDAs), and distributed random scheduling. We also describe example applications of our technique.
5. Proof engineering: Refactoring proof
   Eliot Setzer (University of Edinburgh)
6. Vehicular ad hoc networks
   Eugenio Giordano (University of Bologna)
   

   Show/hide abstract

   Advances in consumer electronics and satellite navigation aremaking vehicular networking a reality. Early applications arelikely to use simple vehicle-to-vehicle orvehicle-to-infrastructure one-hop networking for safety alerts androadside information delivery. However, we envision a plethora ofadvanced applications that may require a clean slate re-design ofthe protocol stack. These networks can be characterized as beinghighly mobile with frequent partitioning, and variable nodedensity. With this poster we present both our simulation results and a description of the UCLA Campus Vehicular Testbed(C-VeT). The simulations yield to a deeper understanding of the behavior of vehicular networks as a function of mobility, and network structure. C-VeTwill provide the research community with aremote-accessible, fully virtualized platform to design, developand evaluate the next generation of vehicular protocols.
7. Automatic derivation of loop bounds
   Florian Zuleger (Technical University Darmstadt)
   

   Show/hide abstract

   In many industries including robotics, consumer electronics, avionics, automotive, and manufacturing, the system components must interact according to a stringent real-time schedule. It is therefore crucial for system engineers to have a good understanding of the worst case execution time (WCET). Recent years have seen a rapid development in automatic termination/ liveness provers, most notably Terminator. The goal of this disseration is to leverage these methods for WCET. Current techniques for termination are not constructive, i.e., they do in general not give an explicit time bound when the program is guaranteed to terminate. It is therefore crucial to extend the mathematical and logical techniques to obtain constructive bounds.
8. Mechanized foundations of finite group theory (handouts)
   Francois Garillot (INRIA)
   

   Show/hide abstract

   We report on a long-term formalisation effort on finite group algebra, aimed at a mechanized proof of the Feit-Thompson theorem. With a better understanding of this particularly long and intricate theorem as a goal, our work devotes considerable attention to the scalability of our proof engineering choices. We use the Coq proof assistant, extended by the SSReflect library to benefit from small scale reflection. Our approach already furnishes a leading development of finite group theory, that showcases the advantages of syntactic notation facilities, as well as type system constructs such as canonical structures.
9. New apparatus for characterising molecular computing substrates
   Gareth Jones (University of Southampton)
   

   Show/hide abstract

   Due to the complexity of molecular computing substrates such as enzymes, characterising them requires vast quantities of data to be obtained from experiments. Data obtained by experimentation is limited by  resource consumption.  Equipment limitations and cost, sample handling precision and human error can lead to excessive consumption in an experiment.  A new method is being developed to resolve these issues and will initially be evaluated in experiments on enzymatic signal processing.
10. Contact network modeling of flu epidemics
    Anilkumar Sorathiya & Ian Leung (University of Cambridge)

    Show/hide abstract

    Epidemic spreading is an intricate biological and sociological process which involves multiscale events from the molecular level of virus-cell interaction to the physical behavioral patterns in a population. Using actual census, family and age structure, land-use and population-mobility data, we have developed a model and related software to study disease propagation. In particular, we address age dependency in epidemic spreading, in which we look into age-dependent locations, contact networks, infection propensities as well as vaccination techniques. As a case study we study influenza epidemics in the UK. The results indicate the relative merits of different vaccination strategies combined with early detection without resorting to mass vaccination of a population.
    Epidemic spreading is an intricate biological and sociological process which involves multiscale events from the molecular level of virus-cell interaction to the physical behavioral patterns in a population. Using actual census, family and age structure, land-use and population-mobility data, we have developed a model and related software to study disease propagation. In particular, we address age dependency in epidemic spreading, in which we look into age-dependent locations, contact networks, infection propensities as well as vaccination techniques. As a case study we study influenza epidemics in the UK. The results indicate the relative merits of different vaccination strategies combined with early detection without resorting to mass vaccination of a population.

Posters Session 3

1. Autonomous experimentation methods for characterising molecular computing substrates
   Chris Lovell (University of Southampton)
   

   Show/hide abstract

   Today’s computing technology is built on a narrow foundation of materials. Nature demonstrates efficient information processing implemented with macromolecular computing substrates. Molecular materials possess a variety of properties that make them attractive for future computing technologies.  However, traditional engineering and design approaches are ill suited to the complexity of these molecules. New methods are required to add macromolecules to the toolkit of the computer engineer. A crucial step towards this goal is the characterisation of the behaviour of macromolecules in the context of other molecules. The present project investigates and develops methods for efficiently characterising molecular materials through computer controlled experimentation. The ultimate aim of this work are algorithms that emulate the experimentation strategies of human experimenters.
2. Dynamic dependency graphs – How much parallelism is out there?
   Jonathan Mak (University of Cambridge)
   

   Show/hide abstract

   We examine the limits of parallelism in existing sequential programs. This is achieved by constructing Dynamic Dependency Graphs out of their execution profiles. The graphs are then analysed to find the critical path and from that the average parallelism. We look at the effect different types of dependencies, in particular true dependencies on the stack pointer, have on parallelism. Our results show that a significant amount of parallelism is available, and provide suggestions as to where this parallelism may be realised.
3. Pan-tropical modelling of the impact of land use change on water quantity provision and low flows
   Jorge Peña Arancibia (King’s College London)
   

   Show/hide abstract

   Land use change, particularly deforestation has the potential to affect the hydrologic cycle. This is of significant importance in areas where precipitation exhibits strong seasonality and spatial variability and water availability during the dry season is necessary to sustain ecosystems and agriculture. Results from observations in small experimental catchments suggest that deforestation increases total water yields. However, these studies cannot be considered representative of natural conditions present in large areas were land clearing and subsequent land use result in severe soil disturbance. Post-forest land cover and soil conditions vary widely in quality and therefore in vegetation water use, infiltration and runoff response to rainfall. In catchments with progressive soil degradation, dry season flows may decrease, irrespective of the increases in total flows. The aim of this study is to provide a pan-tropical assessment of the most sensitive landscapes in terms of hydrological impacts of LUCC incorporating hydrological processes including land use impacts on infiltration, runoff and subsurface flows, evapotranspiration and the best available data at the pan-tropical scale. For this purpose, the FIESTA water resources simulation model, already used throughout Latin America to quantify water resources, will be enhanced to simulate the relevant hydrological processes – at the adequate temporal (monthly) and spatial scale – that govern changes in streamflow regime after land use change. The model and its results will be available in an online GIS-based hydrological suite of models, incorporating new databases including global databases of derived soil hydraulic properties and streamflow validation data. To achieve this modeling across the tropics, significant computational advances will need to be made, modelling system will be converted for use in 64-bits and in a load balanced computing environment. The resulting information will be made available intuitively to decision makers by using recent MS developments in online mapping (virtual earth and virtual earth 3D).
4. The Infinit file system
   Julien Quintard (University of Cambridge)
   

   Show/hide abstract

   As computer networks grow in importance as a medium of exchange of information, users’ need for reliably storing and sharing data increases as well. Peer-to-peer file systems exhibit many interesting properties as a uniform way of accessing information in a completely decentralised way. Unfortunately, such systems are extremely difficult to design due to the inherent byzantine nature of peer-to-peer environments.
5. Designing robots for people
   Laurel Riek (University of Cambridge)
   

   Show/hide abstract

   As robots enter domestic environments in greater numbers it is important that people are able to interact with them in a natural way to increase their acceptance and use. One critical aspect of natural interaction is how emotions are conveyed and understood by a robot.
6. Diagrammatic reasoning across multiple domains: The automatic construction of useful diagrams
   Matthew Ridsdale (University of Cambridge)
7. Noise from in-silico biology: A statistical perspective
   Michele Forlin (CoSBi)
   

   Show/hide abstract

   The study of biological systems through discrete state-space stochastic models is a powerful tool that provides a rigorous conceptual framework for capturing in unambiguous executable format the available information, determining dynamic evolution and predicting the observed behavior of biological systems under diseases scenarios, mutations or drug induced perturbations. It becomes necessary then to develop methods and tools to manage the stochastic simulation results. The work presents two approaches: the first, for multi-run stochastic simulation analysis, is based on Independent Component Analysis (ICA), while the second approach uses statistical time regression models to estimate the evolution in time of the concentration of a species with respect to the others. Both approaches represent a powerful tool to describe the noise effects arising from in-silico biological stochastic simulations.
8. Supporting village community through connected situated displays
   Nick Taylor (University of Lancaster)
   

   Show/hide abstract

   The aim of this project is to investigate the ways that public, digital displays can help support rural communities by improving communications and awareness or fostering a sense of history and identity, and evaluate the techniques we can use while designing these displays. This will be achieved through real world deployments with the participation of residents in a local village to provide feedback and design ideas.
9. Examining the adoption and usage of m-banking applications in low-income communities: The case of M-PESA in Kenya
   Olga Morawczynski (University of Edinburgh)
   

   Show/hide abstract

   This research will ethnographic methods to examine the adoption and usage of M-PESA, a Kenyan m-banking application. In particular, it will analyze how the application is being used by the low-income segment of the population. It will also explain why it is being use in these ways. This analysis will take place in two sites: (1) Kibera, an informal settlement on the outskirts of Nairobi; (2) Bukura, a small village in Western province. This analysis will make clear how the M-PESA application is fitting into already established financial patterns of low-income constituents. It will also make clear how patterns are changing as Kenyans adopt M-PESA.

Posters Session 4

1. Shape grammar and image based 3D reconstruction of buildings
   Olivier Teboul (École Centrale Paris)
   

   Show/hide abstract

   While very topical, creating 3D content is still very challenging. It’s even more tedious when aiming at large scale contents like a whole city. In this work, we tackle grammar-based techniques to generate efficiently large scale urban environment. We will propose a general framework adapted from the work of Muller & al to cope with image-based reconstruction of Paris buildings. Although the optimization method will not be considered in details, some clues are given to understand why such techniques can efficiently represent complex geometry and may solve 3D reconstruction problems.
2. The realities of graphical passwords
   Paul Dunphy (Newcastle University)
   

   Show/hide abstract

   Alphanumeric passwords are a ubiquitous method of knowledge-based authen- tication that can deliver high levels of theoretical security in limiting access to a resource. However the realities of modern day password management requirements, and human cognitive limitations mean this theoretical security is dicult to realise. This problem manifests itself in the form of users choosing guessable passwords, vulnerable to an automated dictionary attack, and other less sophisticated guess attacks from social engineers and individuals with a close relationship to the user. Often such weak passwords are kept for each account the user owns too, meaning one password breach yields access to all password protected services to which the user is enrolled. Graphical passwords are one proposed alternative that aim to address the lack of usability prominent in alphanumeric password systems. Such systems take advantage of the innate human ability to recognise images for purposes of authentication. Current research is geared towards exploring the vast design space of graphical passwords leaving key questions of real world use unanswered. Such questions include the vulnerability of schemes to shoulder sur ng, the extent to which interference oc- curs in memory between multiple graphical passwords, and the ability of users to disclose passwords using description. To date, there has been a shortcoming in user study methodology used to evaluate graphical schemes due to the design lacking ecological validity. Work must focus towards a shift in authentication user study methodology to model scenarios realistically, as this will ultimately become a crucial tool in evaluating performance of graphical passwords in the aforementioned areas, as well as the performance of authentication systems in general.
3. Learning the temperature of a game
   Philipp Hennig (University of Cambridge)
   

   Show/hide abstract

   We attempt to combine the pure mathematical theory of combinatorial games with applied machine learning techniques to develop a parallelizable architecture for approximate game tree search.
4. Automated planning with goal utility dependencies within a satisfiability framework
   Richard Russell (University of Cambridge)
   

   Show/hide abstract

   Some planning problems have many more goals than can be achieved with the available resources. Deciding which goals to plan for is the aim of oversubscription planning. Up until recently, systems have assumed that each goal is equally useful or statically weighted but I have addressed how to incorporate general utility functions over goals. This is aimed at situations where achieving a collection of goals is significantly more useful than the sum of their individual utilities. I am working on incorporating these ideas into a SAT-based planner.
5. From BlenX to SBML
   Roberto Larcher (CoSBi)
   

   Show/hide abstract

   Recently process calculi have been used to model biological systems. The process calculi-based language BlenX is used to develop executable models starting from the description of the molecules involved in the system.  We present here an algorithm to export BlenX programs in SBML format. Starting from a BlenX program, the algorithm extracts the list of molecules defined  in the model and find out reactions these molecules can perform. Then lastly it translates them into an SBML file.  This approach gives us the possibility to share the models we develop in BlenX with the scientific community.
6. Deployment of wireless sensor networks
   Ruoshui Liu (University of Cambridge)
   

   Show/hide abstract

   Wireless Sensor Networks (WSNs) provide a new paradigm for sensing and disseminating information from various physical environments, with the potential to serve many and diverse applications. Successful deployment of WSNs in real environments requires a comprehensive knowledge of the radio channel characteristics within these environments, since the channel has a significant impact on the coverage range and quality of the radio links between nodes. The research investigates the use of frequency and space (antenna) diversity techniques applied to WSNs to ameliorate the impact of the wireless channel on the communication link and improve the robustness of the radio links.
7. Provable security at implementation-level
   Sebastian Faust (Katholieke Universiteit Leuven)
   

   Show/hide abstract

   Traditional provable security treats cryptographic algorithms as black boxes. An adversary may have access to inputs and outputs, but the computation within the box stays secret. Unfortunately, this model often does not match reality where an adversary can attack the algorithm’s implementation with more powerful attacks. An important example in this context are side-channel attacks, which provide an adversary with a partial view on the inner working of hardware. The goal of this project is to develop theoretical models in which formally provable security guarantees can be made concerning the implementation of cryptographic schemes. We studied existing security models that take into account specific attacks on the implementation. These attacks give the adversary either access to additional information through side-channels, or allow it to tamper with the physical device. Unfortunately, it turns out that the existing models fall short: on the one hand they are not suitable to formally analyze theoretical or practical constructions, and on the other the incorporated attacks are of little relevance in practice. In the next phase of this research project, we will try to close this gap and propose new security models that consider less general though more relevant adversaries. Along the lines of the analysis of boolean circuits in the power analysis model seems to be a promising next step. A first analysis already showed that the constructions from Ishai et. al do not provide provable security in this new model.
8. Renewable energy in data centres
   Sherif Akoush (University of Cambridge)
   

   Show/hide abstract

   There are directions to supply power for data centres from renewable energy sources like (wind, solar...). This paper illustrates specific simulations for wind power in UK and how wind intermittency might be solved. It also discusses wind and solar power worldwide and how a simulator could be built to run experiments. Moreover, how can I generalize the load optimizer (balancer/un-balancer) to take into considerations power fluctuations, different workloads and SLA guarantees.
9. Stochastic modelling of single cell assay data
   Simon Youssef (Ludwig-Maximilians-Universität München)
10. Interactive visualization, annotation and processing of terabyte-scale 3D image volumes
    Tahir Mansoori (University of Oxford)
11. A robust video receiver allowing WIMAX video broadcasting and indoor WIFI retransmission
    Usman Ali (Supélec)
    

    Show/hide abstract

    Our current work deals with header recovery at MAC layer of WiMAX. Several redundancies present in the MAC header and also the header check sequences (HCS) that covers only the header, can help to recover the erroneous headers and thus minimizing the possibility of dropping the packet. Thus increasing the possibility of seemingly erroneous video packets reaching the application layer, where one can use tools like JSCD to cope with them. This wok provides a system architecture, where one can apply the proposed header recovery method.

Posters Guidelines

Posters should should be designed for A1 portrait (594 mm x 841 mm) colour printing (either PowerPoint or PDF) and articulate clearly and concisely either visually or textually:

• What challenge is being addressed or question being answered by the research in such a way that a non-expert can understand the importance of the research.
• What the research is.
• What the intended outcome is.
• What stage it is at.
• Any research results, preliminary conclusions, or any potentially exciting or interesting next steps are.

Posters should be aimed at other students and researchers who do not necessarily have expertise in that specific area of research.

Posters should also clearly display your name and the name of your university.

You can find example of posters on the page of the 2007 Summer School.

Discussions Themes

This type of discussions was requested by the students who attended the 2006 Summer School and the students who attended the 2007 Summer School enjoyed them very much. The goal is to make the students think and discuss about a topic or issue of importance and get to know each other.

Each group will be asked to share briefly to the other groups the outcome of their discussion.

Ethics in scientific research

Discussion leader: Andrew Fitzgibbon (Microsoft Research)

Scientific method in computing research

Discussion leader: Jon Crowcroft (University of Cambridge)

Computer science education

Discussion leader: Kevin Bond (Aylesbury Grammar School, Cambridge)

Impact of research

Discussion leader: Wolfgang Emmerich (University College London)

Being a post-doc - Academia vs. industry

Discussion leader: Andrew Phillips (Microsoft Research)

Accommodation

Accommodation will be provided, for non-Cambridge students, at New Hall from Monday 7 July 2007 until the morning of Saturday 12 July.

Group Photo

(Click to view larger image.)

Leisure

Food and Drinks

Cambridge Pubs Guide

Cambridge Restaurants Guide

Theatres

Cambridge Arts Theatre
6, St. Edwards Passage, Cambridge, Cambridgeshire CB2 3PL Tel: 01223 578933   For major theatre productions and national tours.  

ADC Theatre
Park St, Cambridge, Cambridgeshire CB5 8AS Tel: 01223 359547  

Mumford Theatre
Anglia Polytechnic University, East Road, Cambridge, Cambridgeshire  Tel: 0845 196 2320  

Cambridge Corn Exchange
3, Parson Court, Wheeler St, Cambridge, Cambridgeshire CB2 3QE Tel: 01223 357851   For rock, pop, comedy and opera.  

The Junction
Cambridge Leisure Park, Clifton Road, Cambridge, Cambridgeshire CB1 7GX Tel: (01223) 511511 Also for rock, pop, comedy.    

Cinema

The Arts Picture House
38-39, St. Andrews St, Cambridge, Cambridgeshire CB2 3AR Tel: 01223 572929   Arthouse and international.  

Cineworld
Cambridge Leisure Park, Clifton Road Tel. 0871 200 2000  Multi screen Cinema complex.  

Vue Cinema
The Grafton Centre, East Rd, Cambridge, Cambridgeshire CB1 1PS Tel: 0871 2240240   Multi screen Cinema.

Contact Details

If you have any question, send an e-mail message to camevent@microsoft.com

• European Science Initiative

• PhD Scholarship Programme
• 2007 Scholars

• Cambridge University Computer Laboratory
• MSR Summer School 2008 Facebook Group