WISC Seminar: Matt Purver
Reader in Computational Linguistics, School of Electronic Engineering and Computer Science, Queen Mary, University of London
Thursday, 14th January 2016 at 4pm
Venue: CS101, Department of Computer Science
(Coffee and Cake from 3.30pm in the common room)
Title: Language Processing for Diagnosis and Prediction in Mental Health
Abstract: Conditions which affect our mental health often affect the way we use language; and treatment often involves linguistic interaction. This talk will present work on three related projects investigating the use of NLP techniques to help improve diagnosis and treatment for such conditions. We will look at clinical dialogue between patient and doctor or therapist, in cases involving schizophrenia, depression and dementia; in each case, we find that diagnostic information and/or important treatment outcomes are related to observable features of a patient's language and interaction with their conversational partner. We discuss the nature of these phenomena and the suitability and accuracy of NLP techniques for detection and prediction.
WISC Seminar: Suranga Chandratillake
Sales, Marketing, Venture Capital and other adventures in commercial Computer Science
Thursday, 3rd December 2015 at 4pm
Venue: CS101, Department of Computer Science
(Coffee and Cake from 3.30pm in the common room)
Abstract: After a Computer Science degree at Cambridge and a brief flirtation with academia, Suranga ended up joining a succession of start-ups, eventually becoming a CTO and then the founding CEO of his own company in Silicon Valley. After running this company for eight years and learning about alien concepts like marketing, sales and—shudder—corporate finance, he eventually quit to start investing in other peoples’ technology companies. He’s been doing that, as a venture capitalist and angel investor for about four years. In this session, Suranga will share some of the high (and low) lights of taking this non-Computer Science career path and where his degree served him well and where it dismally failed him. There will plenty of time for questions from any who are considering branching out into the more commercial half of our industry.
WISC Seminar: Dr Lambros Lambrinos, University of Cyprus and Imperial College
Smart Cities from a networking and application perspective
Wednesday, 25th November 2015 at 4pm
(Coffee and Cake from 3.30pm in the common room)
Abstract: Applications of the Internet of Things (IoT) are envisaged to transform the urban area landscape. The anticipated large scale deployment of heterogeneous devices generating data creates significant challenges for the underlying networking infrastructure due to the data volume variations and the different delivery reliability and delay requirements of applications. In this talk a number of communication technologies that are prominently used in smart cities will be presented along with upcoming solutions specifically targeting long range communications. The focus will then shift towards public parking management which is a specific smart city application that exploits communication technologies at different levels. This application which is becoming widespread in many cities worldwide will be described from a number of viewpoints: communications, data processing and real-time decision / policy making. The results of an extensive survey on driver requirements will be presented along with a system specifically designed for people with disabilities.
Bio: Lambros Lambrinos has an undergraduate degree from the University of Manchester and a PhD degree in Computer Science from University College London. He holds an Assistant Professor post at Cyprus University of Technology and he is also currently at the ICRI Sustainable Connected Cities Institute at Imperial College London. His research interests evolve around communication networks, ubiquitous computing in general as well as applications of the Internet of Things in urban areas.
WISC Seminar: Dr Ekaterina Kochmar, University of Cambridge
Detecting Learner Errors in the Choice of Content Words Using Compositional Distributional Semantics
Thursday, 22nd October 2015, 4pm,
Venue: CS1.01, Department of Computer Science
(Coffee and Cake from 3.30pm in the common room)
Abstract: With the number of non-native speakers of English growing every year, automated learner error detection and correction has become a popular application area for machine learning algorithms in natural language processing. However, most of the research in this area focuses on function words (e.g., articles and prepositions), while much less attention has been paid to errors in the choice of content words (e.g., nouns, verbs and adjectives). The difference in the nature of the errors in these two groups of words is crucial: since the number of possible confusions and corrections for content words cannot be reduced to any finite set, we cannot easily cast error detection in these combinations as a multi-class classification problem as it is typically done for function words.In this talk, I will present our research on error detection and correction of errors in content words, discuss the challenges and possible solutions for this task. For our research, we have collected a dataset of content word combinations produced by learners of English and annotated them as correct or incorrect. We use methods of compositional distributional semantics and represent content word combinations as vectors in semantic space. Next, we explore the different properties of these vectors that distinguish the semantic representations of the correct combinations from those of the incorrect ones linking our approach to semantic anomaly detection in native English. The error detection task is then cast as a binary classification problem, and the vector properties are used to derive features for a machine learning classifier. Our algorithm yields state-of-the-art results on this task and detects errors with an accuracy of 0.65, with the upper bound based on the human annotators agreement set as 0.75.
Bio: I am a research associate at the Computer Laboratory of the University of Cambridge, working on application of compositional distributional semantics to learner language. My research interests include native language identification, language testing and assessment, error detection and correction, and in particular, I investigate how methods of compositional distributional semantics can be used to detect and correct errors in the choice of content words by language learners. I hold a PhD in Natural Language and Information Processing and an MPhil in Advanced Computer Science from the University of Cambridge.
WISC Seminar: Dr Nigel Collier, University of Cambridge
Exploiting NLP for Digital Disease Informatics
Thursday, 15th October 2015, 4pm,
Venue: CS1.01, Department of Computer Science
(Coffee and Cake from 3.30pm in the common room)
Abstract: Accurate and timely collection of facts from a range of text sources is crucial for supporting the work of experts in detecting and understanding highly complex diseases. In this talk I illustrate several applications using techniques that baseline Natural Language Processing (NLP) pipelines against human-curated biomedical gold standards. (1) In the BioCaster project , high throughput text mining on multilingual news was employed to map infectious disease outbreaks. In order to detect norm violations we show the effectiveness of a range of time series analysis algorithms evaluated against ProMED-mail; (2)In the PhenoMiner project, using an ensemble approach together with SVM learn-to-rank, we show how named entity recognition can achieve improved levels of performance for biomedical concepts. We show however that performance still remains fragile when adapting to new disease domains; (3) Finally, I will discuss how in the SIPHS project we are building concept recognition systems based on deep learning to understand the ‘voice of the patient’ in social media messages.
Bio:Nigel is an EPSRC Fellow and a Principal Research Associate at the University of Cambridge where he is Co-Director of the Language Technology Lab. He graduated from the University of Manchester Institute of Science and Technology (UMIST) in 1996 with a PhD in computational linguistics and has conducted research in both the UK and Japan. Nigel's recent research is focused on bringing together computational techniques such as machine learning, syntactic parsing and concept understanding with the aim of providing a machine-understandable semantic representation of text. This is used to support real-world tasks in the health domain such as disease alerting, conceptual encoding and knowledge discovery.
WISC Seminar: Ben Ward, Oxford Flood Network
The Smart City Built by Citizens
Thursday, 6th October 2015, 4pm,
Venue: CS1.01, Department of Computer Science
(Coffee and Cake from 3.30pm in the common room)
Abstract: In the floodplain of Oxford members of the local community are installing their own water-level monitoring sensors. Inspired by the crowdsourced Japan Radiation Map they are sharing local knowledge about rivers, streams and groundwater to build a better, hyper-local picture of the situation on the ground. Some properties have boreholes which can be used to determine groundwater levels. Some have water sloshing about under their living room in their floor void. And those who live by swollen streams have intimate knowledge of the conditions that lead to floods. These are all great indicators of imminent flooding but are often passed around by word of mouth in a local community. This working demonstrator and reference design forms the basis of a blueprint for communities to build their own sensor networks to highlight their own issues – air quality, radiation, noise, whatever they can find a sensor for.
WISC Seminar: Dr Mirco Musolesi, UCL
Mining Big (and Small) Mobile Data for Social Good
Thursday, 11th June 2015, 4pm, Venue: CS1.04, Department of Computer Science (Coffee and Cake from 3.30pm)
Abstract: An increasing amount of data describing people’s behaviour is collected by means of applications running on smartphones or directly by mobile operators through their cellular infrastructure. This information is extremely valuable for marketing applications, but it has also an incredible potential to benefit society as a whole in various fields, from healthcare to national security. At the same time, mobile data is highly personal. Privacy concerns are increasingly at the center of the public debate and they might potentially hamper the success of this emerging field.
In this talk I will analyze the challenges and opportunities in using big (and small) data for applications of high societal and commercial impact discussing the current work of my lab in the area of mobile data mining. More specifically, I will focus on projects in the area of social media analysis (in particular location-based social networks) and computational epidemiology. I will outline potential privacy issues and practical solutions to address these concerns. The scope of my talk will be broad, encompassing both modelling and systems-oriented issues.
Bio:Mirco Musolesi is a Reader in Data Science at the Department of Geography at UCL. He holds a PhD in Computer Science from UCL and a Master in Electronic Engineering from the University of Bologna. He held research and teaching positions at Dartmouth College, Cambridge, St Andrews and Birmingham. His research interests are highly interdisciplinary and they lie at the interface of different areas, namely large-scale data mining, network science and ubiquitous computing.
DCS Seminar: Prof. Michael W. Marcellin, University of Arizona
Visually-Lossless JPEG2000 for Interactive Multi-Resolution Delivery of Imagery
Wednesday, 10th June 2015, 3pm, Venue: CS1.04, Department of Computer Science (Coffee and Cake session starts from 2.30pm)
Abstract: Visibility thresholds play an important role in finding appropriate quantization step sizes in image and video compression systems. After a brief tutorial on the JPEG2000 standard, we present a method of measuring visibility thresholds for quantization distortion in JPEG2000. A quantization distortion model for each subband is developed based on the statistical characteristics of wavelet coefficients and the dead-zone quantizer of JPEG2000. This is in contrast to previous studies which have assumed uniform quantization distortion. The resulting visibility thresholds are further adjusted for locally changing backgrounds through a visual masking model, and then used to determine the minimum number of coding passes to be included in a JPEG2000 codestream for visually lossless quality. In our experiments, the proposed coding scheme achieves visually lossless coding for 24-bit color images at approximately 20% of the bitrate required for numerically lossless coding.
JPEG2000 inherently supports the display of imagery at various resolutions. When an image is displayed at different resolutions, the spatial frequencies of subbands are changed. Previous JPEG2000 visually lossless algorithms have employed a single set of visibility thresholds optimized for full resolution. This generally results in visually lossless quality at all resolutions, but with significant inefficiencies at less than full resolution. In this talk we discuss a method to minimize the amount of data needed for display at lower resolutions. Specifically, we present a layering strategy which effectively incorporates a different set of visibility thresholds for each resolution. This allows for visually lossless decoding at a variety of resolutions, using only a fraction of the full resolution codestream.
All codestreams produced using the methods described in this talk are fully JPEG2000 Part-I compliant.
Speaker's bio: Michael W. Marcellin graduated summa cum laude with the B.S. degree in Electrical Engineering from San Diego State University in 1983, where he was named the most outstanding student in the College of Engineering. He received the M.S. and Ph.D. degrees in Electrical Engineering from Texas A&M University in 1985 and 1987, respectively.
Since 1988, Dr. Marcellin has been with the University of Arizona, where he holds the title of Regents' Professor of Electrical and Computer Engineering, and of Optical Sciences. His research interests include digital communication and data storage systems, data compression, and signal processing. He has authored or coauthored more than two hundred publications in these areas.
Dr. Marcellin is a major contributor to JPEG2000, the emerging second-generation standard for image compression. Throughout the standardization process, he chaired the JPEG2000 Verification Model Ad Hoc Group, which was responsible for the software implementation and documentation of the JPEG2000 algorithm. He is coauthor of the book, JPEG2000: Image compression fundamentals, standards and practice, Kluwer Academic Publishers, 2002. This book serves as a graduate level textbook on image compression fundamentals, as well as the definitive reference on JPEG2000. Dr. Marcellin served as a consultant to Digital Cinema Initiatives (DCI), a consortium of Hollywood studios, on the development of the JPEG2000 profiles for digital cinema.
Professor Marcellin is a Fellow of the IEEE, and is a member of Tau Beta Pi, Eta Kappa Nu, and Phi Kappa Phi. He is a 1992 recipient of the National Science Foundation Young Investigator Award, and a corecipient of the 1993 IEEE Signal Processing Society Senior (Best Paper) Award. He has received teaching awards from NTU (1990, 2001), IEEE/Eta Kappa Nu student sections (1997), and the University of Arizona College of Engineering (2000, 2010). In 2003, he was named the San Diego State University Distinguished Engineering Alumnus. Professor Marcellin is the recipient of the 2006 University of Arizona Technology Innovation Award. He was finalist for the 2012 Arizona Governor’s Innovation Awards. From 2001 to 2006, Dr. Marcellin was the Litton Industries John M. Leonis Distinguished Professor of Engineering. He is currently the International Foundation for Telemetering Chaired Professor of Electrical and Computer Engineering at the University of Arizona.
DCS Seminar: Dominic Orchard, Research Associate at Imperial College
Effects as sessions, sessions as effects
Thursday 21st May 2015, 4pm, Venue: CS1.01, Department of Computer Science (Coffee and Cake session starts from 3.30pm)
Abstract: The lambda calculus and pi-calculus are two useful, well-known models of computation. Milner showed that the pi calculus subsumes the
expressivity of the lambda calculus by giving a translation of lambda terms into pi calculus processes. Recently it has been shown that a typed translation is also possible: the simply-typed lambda calculus can be translated into the pi-calculus augmented with the session-typing discipline. That is, the type theory of the lambda calculus is subsumed by session types, which describe and restrict the communication behaviour of pi-calculus programs.
But session types are even more powerful than this. In this talk, I show that session types also subsume effect systems. This is shown by an effect-preserving translation from the lambda calculus with a type-and-effect system (for example, for reasoning about state side effects) into the session-typed pi-calculus; effect types are represented as session types. This expressivity result can be leveraged, for example, to safely introduce implicit parallelism when compiling functional code, even in the presence of possible side-effects.
A natural question is then: can we go back the other way? Can session types be encoded into a lambda calculus with an effect system? I'll describe preliminary results which suggest that this indeed possible with a translation of session-typed processes into effect-typed functions. This can be used to embed session types into the type systems of existing functional languages, e.g., Haskell, which I will demonstrate.
This is joint work with Nobuko Yoshida.
Bio of the speaker: Dominic is a research associate at Imperial College, London in the Mobility Reading Group. He was previously
a post-doc at the University of Cambridge where he also completed his PhD in the area of semantics and type
systems for contextual computation. He graduated with an MEng in Computer Science from the University of
Warwick in 2008.
WISC Seminar: Peter Coetzee, Phd candidate, University of Warwick
Chiron: Data Intensive Computing for Warwick
Wednesday 20th May 2015 11am, Venue: D1.07, Zeeman building
Abstract: In support of CDTs in Urban Science and Mathematics for Real World Systems, as well as the wider research community at Warwick, the University is in the final stages of delivery for a unique heterogeneous data-intensive compute cluster, named Chiron. This talk will cover the four key capabilities in Chiron: Hadoop for offline compute; streaming analytics; hardware accelerated analytics research platform; and a high-memory compute capability, as well as how these four capabilities may be used in harmony. Programming models, tools, and practical advice for making the most of the machine will be discussed as well. If you are planning to work with large datasets in your research, and want to better understand some of the latest resources available to researchers at the University, this should give you the jumping-off point to start preparing your codes and data for ingest into Chiron once it is generally available later this year!
DCS Seminar: Yosi Keller, Associate Professor, Faculty of Engineering, Bar Ilan University, Israel
Biometric features estimation from facial images
Monday 18th May 2015, 3pm, Venue: CS1.04, Department of Computer Science
Abstract: In this work we propose two novel approaches for estimating biometric features based on face images. We estimate age, race and gender, but focus mainly on age estimation. Our first approach improves the age estimation by formulating the Diffusion Framework as a boosted supervised learning algorithm, where at each step we estimate a subset of the biometric attributes that are used to augment the feature space in the succeeding estimation step. The feature spaces are embedded in the Diffusion embedding space to yield adaptive bases that are used for age regression. In our second contribution we propose to represent the set of face image by a deep tree, partitioned according to biometric attributes and coarse age
estimates. Thus, the image data is in each graph leaf relates to simpler manifold geometries and can be estimated separately. The tree construction and regression are based on Kernel SVM regressors and we show that their accuracy can be significantly improved by applying low rank distance learning. We tested the resulting schemes on the state-of-the-art MORPH-II dataset, and showed a major improvement compared to contemporary works.
Bio of the speaker: Yosi Keller is an Associate Professor with the Faculty of Engineering, Bar Ilan University, Ramat Gan, Israel. He is the director of the Data Analysis research group at the Faculty of Engineering. Prof. Keller graduated his PhD studies summa cum laude at Tel Aviv University 2003, and then spent three years as a Gibbs Assistant Professor at the Applied Math group in Yale University, New Haven, Connecticut. In October 2007 he joined the Faculty of Engineering in Bar Ilan University as a Senior Lecturer (the equivalent of Assistant Professor) and was appointed an Associate Professor in April 2012. His work was published in journals such as IEEE TIP, TPAMI and TSP, and he is an Associate editor of IEEE Transactions on Circuits and Systems for Video Technology.
As for industrial affiliations, in 1994-1998 Professor Keller was an algorithms developer and team leader in the Israeli Ministry of Defence R&D department. He later served as the algorithms team leader in Emblaze Ltd. and Infowrap Systems Ltd. Since 2007 he has been a consultant to Israeli and International technology-oriented companies, in the fields of computer vision, machine learning and biometrics.
His recent academic work relates to security and biometric research. In particular, probabilistic multicue object tracking in video, and geometric fingerprints recognition based on spectral graph matching. Geometric object recognition is an ongoing research effort and will be applied towards deriving non-frontal face recognition schemes.
DCS Seminar: James Sumner, Lecturer, University of Manchester
A clichéd history of computing
Thursday 14th May 2015, 4pm, Venue: CS1.04, Department of Computer Science
Abstract: My presentation surveys how electronic computers were presented to, and interpreted by, non-specialist audiences, from their origins in the 1940s to the consolidation of mass public use in the 1990s. I focus particularly on how a small set of understandings, assumptions and explanatory approaches became clichés of the form, circulating and recirculating through educational and general-interest literature. Some clichés had origins in wider public debates, such as the longstanding fear of unemployment and deskilling; others were promoted by the industry itself, notably the GIGO (“garbage in, garbage out”) principle which positioned the technology as a neutral instrument. Some were short-lived, such as Leon Bagrit’s “automation” crusade of the 1960s, whereas some persisted for half a century, such as the remorseless tendency to explain binary arithmetic to all audiences for all purposes. I will consider how much of this cliché reproduction was due to conscious shared aims, and how much to simple convention, and will also offer the beginnings of an attempt to evaluate its influence, via policy determinations, on real-world change.
Bio of the speaker: James Sumner is Lecturer in History of Technology at the University of Manchester. His research in the history of computing usually focuses on British cultural contexts, and has included work on the early promotional rhetoric of British computer manufacturers and the assimilation of the 1980s IBM PC-compatible platform. He has a strong interest in engaging with public audiences about historical research (for various values of “public”), and is currently involved in co-supervising a project with colleagues at the Museum of Science and Industry on the museum presentation of historic sites.
DCS Seminar: Ioannis Psaras, University College London
Overview, Challenges and Prospects of the Information-Centric Networking Paradigm
Thursday 23rd April 2015, 4pm, Venue: CS1.01, Department of Computer Science
Abstract: In this talk we are going to introduce the main concepts of Information-Centric Networks (ICNs), a new and exciting research area with a lot of potential to transform the way we do networking today. We will go into details on the architectural challenges of deploying such an architecture over the current Internet and we will point to interesting directions for future research. We will finally present the In-Network Resource Pooling Principle (ACM HotNets 2014), a congestion control scheme for Information-Centric Networks which presents huge potential in dealing with the daunting problem of congestion in the Internet.
In particular, we will question the widely adopted view of in-network caches acting as temporary storage for the most popular content in Information-Centric Networks. Instead, we will propose that in-network storage is used as a place of temporary custody for incoming content in a store and forward manner. Given this functionality of in-network storage, senders push content into the network in an open-loop manner to take advantage of underutilised links. When content hits the bottleneck link it gets re-routed through alternative uncongested paths. If alternative paths do not exist, incoming content is temporarily stored in in-network caches, while the system enters a closed-loop, back-pressure mode of operation to avoid congestive collapse.
Bio of the speaker: Ioannis Psaras is an EPSRC Fellow at the Electrical and Electronic Engineering Department of UCL. He is interested in resource management techniques for current and future networking architectures with particular focus on routing, replication, caching and congestion control. Before joining UCL in 2010, he held positions at the University of Surrey, and Democritus University of Thrace, Greece, where he also obtained his PhD in 2008. He has held research intern positions at DoCoMo Eurolabs and Ericsson Eurolabs. He is involved and running several EU and EPSRC projects and participates in the technical programme committees of tens of high-quality conferences. More details can be found at:http://www.ee.ucl.ac.uk/~uceeips/ and more discussion/interest is welcome at i.psaras *at*ucl.ac.uk.
WISC Seminar: Dr Vasileios Lampos, Research Associate at University College London and Visiting Researcher at Google
Mining the Social Web
Tuesday 10th March 2015, 4pm, Venue: A401, Department of Engineering
Abstract: A series of statistical NLP case studies Over the past few years user-generated content has been the centre of various research efforts in the domain of statistical natural language processing. In particular, the open nature of the microblogging platform of Twitter provided the opportunity for various appealing ideas to be evaluated. Based on the hypothesis that this online stream of content should represent at least a biased fraction of real-world situations or opinions, we have proposed core algorithms for nowcasting the rate of an infectious disease, such as influenza, or even a natural phenomenon like rainfall rates [1,2]. A simplified emotion analysis on a longitudinal set of tweets revealed interesting patterns, including signs of rising anger and fear before the UK riots in August, 2011 . By extending linear text regression approaches to embed user relevance, we proposed a family of bilinear regularised regression models, which found application in the approximation of voting intention trends . Finally, we attempted to reverse the previous modelling principle to look into how various user attributes or behaviours may influence a generic notion of user-impact .
1. Lampos and Cristianini. Tracking the flu pandemic by monitoring the Social Web, Cognitive Information Processing ‘10.
2. Lampos and Cristianini. Nowcasting Events from the Social Web with Statistical Learning, ACM TIST (2012).
3. Lansdall-Welfare, Lampos and Cristianini. Effects of the Recession on Public Mood in the UK, WWW ‘12.
4. Lampos, Preoţiuc-Pietro and Cohn. A user-centric model of voting intention from Social Media, ACL ‘13.
5. Lampos, Aletras, Preoţiuc-Pietro and Cohn. Predicting and Characterising User Impact on Twitter, EACL ‘14.
DCS Seminar: Dr. Francesc Auli-Llinas, Department of Information and Communications Engineering, Universitat Autònoma de Barcelona, Spain
Image coding via SIMD computing
Thursday 5th March 2015, 4pm, CS101, Department of Computer Science
Abstract: Recent years have seen the upraising of a new type of processors strongly relying on the Single Instruction, Multiple Data (SIMD) architectural principle. The main idea behind SIMD computing is to apply a flow of instructions to multiple pieces of data in parallel and synchronously. This permits the execution of thousands of operations in parallel, achieving higher computational performance than with traditional Multiple Instruction, Multiple Data (MIMD) architectures. Among others, SIMD processing has been employed in fields such as bioinformatics, physics, or analytics to accelerate myriad applications. The results achieved are staggering. In the field of image and video coding, SIMD processing has not achieved its full potential due to the use of algorithms that process the image samples sequentially. This talk will describe a research line carried out in our group in which all the stages of a wavelet-based image coding system are rethought to allow parallel processing of the image samples. The execution of the proposed algorithms in SIMD processors such as GPUs accelerates the coding of images/videos enormously.
Speaker: Francesc Auli-Llinas (http://www.deic.uab.cat/~francesc) is a Ramon y Cajal Fellow with the Department of Information and Communications Engineering, Universitat Autònoma de Barcelona, Spain, where he received the Ph.D. (cum laude) in computer science in 2006. Since 2002 he has been funded with competitive fellowships from the Spanish and Catalan Governments. He carried out two research stages of one year each with Prof. David Taubman, at the University of New South Wales, and Prof. Michael Marcellin, at the University of Arizona. In 2013, he received a distinguished R-Letter given by the IEEE Communications Society for a paper co-authored with Prof. Marcellin. He is reviewer for various magazines and symposiums and has authored numerous papers in journals and conferences. His research interests lie in the area of image and video coding, computing, and transmission.
WISC Seminar: Prof Sophia Ananiadou, National Centre for Text Mining, University of Manchester
Events and pathways: bridging the gap between text and knowledge.
Monday 2nd March 2015, 4pm, Wolfson Research Exchange, Library
Abstract: Recently text mining methods have focused on the extraction of complex events to support a number of biomedical applications. In my talk, I will discuss recent developments for the understanding of cancer mechanisms using named entity recognition and event extraction and their mapping to reactions and their context in pathways. Using an interoperable text mining infrastructure we can easily customise text mining workflows to support the linking between knowledge and text.
WISC Seminar: Dr Niki Trigoni, Associate Professor, Department of Computer Science, University of Oxford
Enabling location-based services for smart cities
Thursday 26th February 2015, 4pm, CS1.01
Abstract:While GPS has become the de facto standard as an outdoor positioning infrastructure, researchers are still actively working on the problem of indoor positioning. This is an exciting area with many practical applications for smart cities, ranging from indoor navigation to location-based content delivery. In this talk, I will present three key problems that challenge the adoption and widespread use of indoor positioning systems. First, I will discuss the problem of non-line-of-sight signal propagation in cluttered indoor environments, and explain how it adversely affects the accuracy of range-based localisation techniques. Second, I will explain the difficulty in positioning nodes in indoor spaces with and without bespoke positioning infrastructure, and will demonstrate the merit of exploiting node encounters (wireless contacts) and map constraints to improve location accuracy. Finally, I will talk about the challenge of positioning nodes in indoor spaces with dense sensor infrastructure, and will present a novel learning approach to assessing the accuracy of co-located positioning systems.
Bio: Niki Trigoni is an Associate Professor at the Oxford University Department of Computer Science and a fellow of Kellogg College. She obtained her PhD at the University of Cambridge (2001), became a postdoctoral researcher at Cornell University (2002-2004), and a Lecturer at Birkbeck College (2004-2007). Since she moved to Oxford in 2007, she established the Sensor Networks Group, and has conducted research in communication, localization and in-network processing algorithms for sensor networks. Her recent and ongoing projects span a wide variety of sensor network applications, with increasing focus on indoor positioning. She has co-authored 75 peer-reviewed papers, and has recently received 4 best paper/poster awards on papers related to positioning algorithms.
DCS Seminar: Dr Madalina Croitoru,Associate Professor, University of Montpellier II, INRIA GraphIK Research team
Reasoning about Knowledge as Graphs: Practical Artificial Intelligence Applications
Thursday 19th February 2015, 4pm, CS1.01
Abstract: During and since my PhD and throughout my postdoctoral and lectureship I have investigated graph based knowledge representation and reasoning languages applied to different Artificial Intelligence fields. Syntactically the objects I am interested in are graphs. The graphs are endowed with semantics sound and complete with respect to syntactic graph operations. I will detail work done in the field of Conceptual Graphs and argumentation graphs. I will also present our work within inconsistency tolerant reasoning and present some practical applications (mainly centered around agronomy) that have benefitted from the formalisms described above.
Bio: Dr. Madalina Croitoru completed her PhD with the Department of Computing Science, University of Aberdeen. Her research looked at improving Conceptual Graph applicability in Artificial Intelligence. She then worked as Research Fellow for the Department of Electronics and Computer Science, University of Southampton involved, part time, in two EU projects: HealthAgents and Open Knowledge. From September 2008 she joined University Montpellier II, France as an Associate Professor and is currently a member of the INRIA research team GraphIK at LIRMM.
WISC Seminar: Prof Peter Elias, Institute for Employment Research, University of Warwick
New forms of data – new opportunities for research?
Wednesday 11th February 2015, 4pm, MOAC Seminar Room, second floor, Senate House
Many new types of data are now seen as potential research resources, but what are they, where are they and how can they be made available for research? Who controls access and what are the rules and regulations regarding access to and linkage between different types of data? This talk outlines the efforts that the UK research councils are making to improve access to and facilitate linkage between different types of data, together with an overview of similar developments at the international level.
Peter Elias is an advisor to the ESRC for the strategic development of data and related resources. He is Deputy Chair of the Administrative Data Research Network Board of the UK Statistics Authority and Deputy Director of Life Study, the newest and largest birth cohort ever conducted in the UK.
DCS Seminar: Dr Hannah Dee, Senior Lecturer, Department of Computer Science, Aberystwyth University
Watching plants grow: computer vision for plant phenomics
Thursday 12th February 2015, 4pm, CS1.01
The global population is growing, but our food supply isn't growing as fast. Plant genetics and intelligent breeding practices have led to massive increases in yield (the so-called "Green revolution") but progress has slowed considerably in recent years. Phenomics – large scale measurement of plant traits – is the bottleneck here, and computer vision is ideally placed to help. This talk will showcase some recent computer vision work in plant imaging, which has the potential to address some of the main challenges: high throughput, non-interactive extraction of biological traits (senescence, biomass, growth) from images of plants.
DCS Seminar: Prof Ursula Martin,Department of Computer Science, University of Oxford
Mathematical practice, crowdsourcing, and social machines
Thursday 5th February 2015, 4pm, CS1.01
Mathematical practice is an emerging interdisciplinary field which draws on philosophy, social science and ethnography, and the input of mathematicians themselves, to understand how mathematics is produced. Online mathematical activity provides a rich source of data for empirical investigation of mathematical practice - for example the community question answering system mathoverflow contains around 40,000 mathematical conversations, and polymath collaborations provide transcripts of the process of discovering proofs. Such investigations show the importance of "soft" aspects such as analogy and creativity, alongside formal deduction, in the production of mathematics, and give us new ways to think about the possible complementary roles of people and machines in creating new mathematical knowledge. Social machines are new paradigm, identified by Berners-Lee, for viewing a combination of people and computers as a single problem-solving entity, and the subject of major international research endeavours. I present progress on a research agenda for mathematics social machines, a combination of people, computers, and mathematical archives to create and apply mathematics, which is being pursued at Oxford under an EPSRC Fellowship.
WISC Seminar: Dr Miles Osborne, Senior Research Scientist, Bloomberg LP
Wall Street meets the Biebs: Finance and Event Detection in Twitter
Monday 2nd February 2015, 4pm, MOAC Seminar Room, second floor, Senate House
The Financial Community has started to embrace Social Media and in particular Twitter. Thompson-Reuters and Bloomberg include tweets in their trading platforms, whilst companies are now allowed to make reports on Facebook and Twitter. Numerous start-ups revolve around identifying financially-relevant posts. Tweets are now big business. Financial Markets are known to move on the basis of news. But do they move when that news appears in Twitter? I examine a broad range of market moving news appearing in traditional newswire and in Twitter, considering financial sector, coverage, latency and just who does this reporting. Results suggest that Twitter can break financially-interesting news before newswire and that this mainly affects commodities. Along the way I will explain how event detection can operate at scale and will outline algorithmic and modelling challenges.
WISC Seminar: Dr Kalina Bontcheva, Senior Researcher, Department of Computer Science, University of Sheffield
Named Entity Linking and Semantic Search using Linked Open Data
Wednesday 21st January 2015, 4pm, MOAC Seminar Room, second floor, Senate House
Abstract: This talk will start by introducing semantic search, named entity linking, and linked open data. Challenges and relevant datasets will be discussed, with particular focus on social media. Next a state-of-the-art approach to named entity linking will be presented. This will be followed by a demonstration of how the enriched texts can be searched semantically, in combination with knowledge from DBpedia, in order to answer queries such as "Documents mentioning politicians born in Sheffield". The talk will conclude by discussing different practical applications of this research.
WISC Seminar: Dr Benjamin Hennig, Researcher, Schoold of Geography and the Environment, University of Oxford
Mapping London - A new approach to visualising cities
Thursday 29th January 2015, 4pm, CS1.01
Abstract: London is a diverse divided city which is often discussed and which with the growing amount of openly available data is increasingly also revealed in cartographic and other visualisations. As part of this, the Londonmapper project builds is the first attempt since the 1970s to create a comprehensive Social Atlas of the city making use of the most recent techniques and most advanced ways of diseminating the results. The data in the new (and growing) online atlas are not mapped in a conventional way, but use different types of cartograms to create a novel picture of London's complex diversity. What we see in these maps and statistics is the fragmented nature of London which is not only a tale of two cities, but a patchwork of divisions that go through many parts of the city in all areas. It’s novel approach helps to see the city in a new light and to raise awareness for the conditions that 21st century Londoners are living in.
WISC Seminar: Malkiat Thiarai, Head of Corporate Information Management, Birmingham City Council
Using Data and Information to improve services to vulnerable and at risk children.
Thursday 4 December 2014, 4pm, CS1.01
Abstract: To support the needs of children who are or may become vulnerable or at risk, there is a growing move to consider and design service provision from the viewpoint of the child rather than the provider. This brings into focus the number of different services who may be in contact with the child and family members and have an influence, direct or indirect, on a child’s experiences as they grow up. This re-design of services challenges existing structures and service provision models that many public sector organisations have and requires new models that have the flexibility to adapt to the city or locality context in which they operate. Integral to creating new models of service delivery focused around children and young people is to understand how information and data flows as a child interacts within its family and social environment. This, in turn, will help inform the way in which children and young people’s interact with universal or targeted service and society as well as the importance of personal and private information being shared.
Prof Rob Stewart and Dr Stephani Hatch, King's College London
Understanding Mental Health in Urban Environments - The south London Experience
Thursday 27 November 2014, 4pm, CS1.01
Professor Rob Stewart and Dr Stephani Hatch will demonstrate the use of multiple complementary methods to examine mental health in urban contexts, drawing on data from a large community survey (SELCoH) and mental health electronic record data resource (CRIS) covering an area of south London with high levels of social, ethnic and economic diversity.
Matt Purver, Queen Mary University London
You Can Tell A Lot About People From What They Say
Thursday 13 November 2014, 4pm, CS1.01
This talk will describe two recent projects investigating the use of computational linguistic methods to discover characteristics of text-based systems from their language. In one, we collected and analysed Twitter messages mentioning the Barbican arts centre, to examine whether people's language segmented into different areas of interest or artistic genres, and discover "boundary crossers" - people who might be interested in more than one genre. In the other, we analysed text from clinical doctor-patient interactions, to investigate what features of people's talk might indicate severity of symptoms and predict progress in various mental health conditions. In both cases, useful insights can be gained, although success depends on choice and robustness of language processing methods.
Norman Poh, University of Surrey
System Design and Performance Assessment: A Biometric Menagerie Perspective
Thursday 6 November 2014, 4pm, CS1.01
One of the major sources of variability in assessing the performance of a biometric system is the subject variability. Testing the same system on two disjoint populations of users almost always invariably yields two different results. This phenomenon was first described by Doddington et al when they evaluated speaker recognition systems in 1998. The users, or more precisely the statistical models associated with them, who are difficult to be recognised by a biometric system are given names such as goats and lambs; whereas users whose biometric traits are likely to be successful at impersonating others are called wolves. This phenomenon is dubbed “Doddington Zoo” and “Biometric Menagerie”. Subsequent studies then either aim at characterising the phenomenon, or at reducing the phenomenon that leads to better tailored decisions such as user-specific decision threshold and user-specific score calibration, and fusion strategies.
In this tutorial, we will describe biometric menagerie and explain how and why it has a direct impact on how the system performance is characterised; how confidence intervals can be estimated; and why performance prediction is difficult.
The significance of biometric menagerie is that it has impact on all biometric modalities. Furthermore, by reducing the phenomenon through user-specific strategies, a relative performance gain of about 30% has been observed for a unimodal biometric system; and up to 50% for a multimodal biometric system.
Hassan Mansour, Mitsubishi Electric Research Laboratories
Factorized Robust Matrix Completion
Tuesday 4th November 2014, 3pm, CS1.01
Recent SVD-free matrix factorization formulations have enabled rank minimization for systems with millions of rows and columns, paving the way for matrix completion for extremely high dimensional data. In this talk, we discuss a robust matrix completion algorithm that uses factorized matrix decomposition with a pre-specified rank to compute the low rank component. We then adopt a block coordinate descent approach using spectral projected gradient steps to alternate between the solution of the low rank component and the sparse component all the while traversing the updated Pareto curve of each subproblem. We illustrate the performance of our algorithm for video background subtraction.
Dr Hassan Mansour is a member of the research staff in the Multimedia Group at Mitsubishi Electric Research Laboratories. He received his M.A.Sc. (2005) and Ph.D. (2009) from the Department of Electrical and Computer, University of British Columbia (UBC), Vancouver, Canada where he conducted research on scalable video coding and transmission. He then pursued a postdoctoral fellowship in the Departments of Mathematics, Computer Science, and Earth and Ocean Sciences at UBC working on theoretical and algorithmic aspects of compressed sensing and its application to seismic imaging.
Stefan Dziembowski, University of Warsaw
Bitcoin contracts---digital economy without lawyers?
Thursday 30 October 2014, 4pm, CS1.01
BitCoin is a digital currency system introduced in 2008 by an anonymous developer using a pseudonym "Satoshi Nakamoto". Despite of its mysterious origins, Bitcoin became the first cryptographic currency that got widely adopted --- as of May 2014 the Bitcoin capitalization is over 5 bln euro. Bitcoin owes its popularity mostly to the fact that it has no central authority, the transaction fees are very low, and the amount of coins in the circulation is restricted, which in particular means that nobody can "print" money to generate inflation. The financial transactions between the participants are published on a public ledger maintained jointly by the users of the system.
One of the very interesting, but slightly less known, features of the Bitcoin is the fact that it allows for more complicated "transactions" than the simple money transfers between the participants: very informally, in Bitcoin it is possible to "deposit" some amount of money in such a way that it can be claimed only under certain conditions. These conditions are written in the form of the "Bitcoin scripts" and in particular may involve some timing constrains. This property allows to create the so-called "contracts", where a number of mutually-distrusting parties engage in a Bitcoin-based protocol to jointly perform some task. The security of the protocol is guaranteed purely by the properties of the Bitcoin, and no additional trust assumptions are needed. This Bitcoin feature can have several applications in the digital economy, like creating the assurance contracts, the escrow and dispute mediation, the rapid micropayments, the multiparty lotteries.
In this talk I will give a short introduction to this area, present some recent results, and highlight the future research directions.#
Francois Grey, Citizen Cyberscience Centre, University of Geneva and Lifelong Learning Lab, Tsinghua University
LEGO2NANO: Designing participative science for children in Chinese cities
Thursday 23rd October 2014, 4pm, CS1.01
China is in the process of reforming its education system. As part of this reform, we envisage a future in which Chinese children can participate actively in scientific research by building low-cost scientific instruments, using them to study their urban environment and sharing the resulting data online for collaborative analysis with crowdsourcing technologies. We have made a first step towards this ambitious goal by launching an international challenge called LEGO2NANO involving collaborations between post-graduate, undergraduate, high school and middle school students. We describe how students involved in the challenge have designed, built and operated a low-cost atomic force microscope to study air pollution in Beijing. We outline next steps, which involve establishing a facility for collaborative design and prototyping of new low-cost scientific instruments suitable for use in Chinese schools, and a virtual lab on the Web where children can share and analyze their results interactively.
Jon Crowcroft, University of Cambridge
Can we build a Europe-only cloud, and should we?
Thursday 9th October 2014, 4pm, CS1.01
Recent events have caused alarm in parts of Europe, both amongst citizens and also in government circles. The Germans, amongst others, have proposed a BundesCloud (or more generally a local-only Cloud, and Internet). In this talk, I will look at what steps are necessary to implement such a thing, if at all possible, and how users would have assurance about its properties. I will also touch on whether (in the light of alternative approaches to privacy) it is really what is wanted.
Ian Parberry, University of North Texas
Generating Realistic Terrain for Video Games using Digital Geography
Thursday 26th June 2014, 4pm, CS1.01
Video games such as Minecraft take place in artificially generated random terrain created using fractal noise generators such as Perlin noise. Unfortunately, since Perlin noise has a uniform gradient distribution, the terrain has a recognizable lumpy-bumpy sameness. We will see the results of a fractal analysis of the gradient distribution at 5 meter resolution in a 160,000 square kilometer region in central Utah, and show how to modify the Perlin noise algorithm to incorporate them. The presentation will end with a slideshow of random terrain of different types generated using this new algorithm and rendered with Terragen 3, a professional-quality ray-casting renderer used in popular movies, TV, advertising, print publishing, and video games.
Laurence Tratt, Software Development Team, King's College London
Towards Language Composition
Thursday 19th June 2014, 4pm, CS1.01
We want better programming languages, but "better" invariably ends up becoming "bigger". Since we can't keep making our languages bigger, what alternatives do we have? In this talk, I propose language composition as a possible solution to this long standing problem. Language composition means merging two languages and allowing them to be used together. At its most fine-grained, this could allow multiple programming languages to be used together within a single source file.
However, language composition is not a new idea. It has failed in the past because editing composed programs was intolerably difficult and the resulting programs ran too slow to be usable. Without good solutions to these problems, language composition will remain an unrealised ideal.
In this talk, I will show how the work we are doing in the Software Development Team is beginning to address both aspects. We have built a prototype editor utilising a novel concept 'language boxes', which allows one to edit composed programs in a natural way, without the limitations of traditional approaches. We are tackling the performance problem by composing together interpreters using meta-tracing, allowing us to build composed VMs with custom JITs that naturally optimise across different language's run-times. While we are much nearer the beginning of the journey than the end, our initial research has allowed us to build a simple composition of two very different languages: Python and Prolog. Joint work with Edd Barrett, Carl Friedrich Bolz, Lukas Diekmann, and Krishnan Vasudevan. More details at http://soft-dev.org/.
Paul Woolman, NHS Forth Valley
Current topics in healthcare information
Thursday 5th June 2014, 4pm, CS1.01
This talk will start with a summary of past and current strategies for healthcare architectures and information systems in the UK and EU. It will then illustrate the current hot topics in R&D and strategic developments in the domain with a few specific projects as illustration. The talk will finish with an outline of the future directions.
Mark Ryan, University of Birmingham
Balancing Societal Security and Individual Privacy: Accountable Escrow System
Thursday 29th May 2014, 4pm, CS1.01
Individual privacy is a core human need, but society sometimes has the requirement to do targetted, proportionate investigations in order to provide security. To reconcile individual privacy and societal security, we explore whether we can have surveillance in a form that is verifiably accountable to citizens. This means that citizens get verifiable proofs of how much surveillance actually takes place. This is joint work with Jia Liu and Liqun Chen.
Stephen Clark, University of Cambridge
Compositional and Distributional Models of Meaning for Natural Language
Thursday 15th May 2014, 4pm, CS1.01
There have been two main approaches to modelling the meaning of language in Computational Linguistics and Natural Language Processing. The first, the compositional approach, is based on classical ideas from Philosophy and Mathematical Logic, and implements the ideas from Formal Semantics which followed Montague's work in the 1970s. The second, more recent approach focuses on the meanings of the words themselves. This is the distributional approach to lexical semantics and is based on the ideas of structural linguists such as Firth, and is sometimes related to the Wittgensteinian philosophy of ``meaning as use". The idea is that the meanings of words can be determined by considering the contexts in which words appear in text. In this talk I will describe describe a new problem in natural language semantics, which is to combine the two approaches to modelling meaning described above, in order to produce meaning representations which are robust and flexible, and yet still respect the structure inherent in sentences. Such representations could revolutionize language processing applications such as "semantic search" and question answering.
Kalina Bontcheva, University of Sheffield
Natural Language Processing (NLP) of Social Media: Are We There Yet?
Thursday 8th May 2014, 2pm, CS1.01
Social media poses three major computational challenges, dubbed by Gartner the 3Vs of big data: volume, velocity, and variety. NLP methods, in particular, face further difficulties arising from the short, noisy, and strongly contextualised nature of social media. As a result, NLP methods generally tend to perform significantly worse on social media, than on longer, cleaner texts, such as news. After a short introduction to the challenges of processing social media, I will outline key NLP algorithms adapted to processing social content, discuss available annotated datasets and outline remaining challenges. Frontier issues such as rumour detection and user geo-location. I will also discuss briefly practical and ethical considerations, arising from gathering and mining social media content.
Sebastian Riedel, University College London
Relation Extraction with Matrix Factorization and Universal Schemas
Thursday 1st May 2014, 4pm, CS1.01
The ambiguity and variability of language makes it difficult for computers to analyse, mine, and base decisions on. This has motivated machine reading: automatically converting text into semantic representations. At the heart of machine reading is relation extraction: predicting relations between entities, such as employeeOf(Person,Company). Machine learning approaches to this task require either manual annotation or, for distant supervision, existing databases of the sameschema (=set of relations). Yet, for many interesting questions (who criticised whom?) pre-existing databases and schemas are insufficient. For example, there is no critized(Person,Person) relation in Freebase. Moreover, the incomplete nature of any schema severely limits any global reasoning we could use to improve our extractions. The need pre-existing datasets can be avoided by using, what we call, a "universal schema": the union of all involved schemas (surface form predicates such as "X-was-criticized-by-Y" as in OpenIE, and relations in the schemas of pre-existing databases). This extended schema allows us to answer new questions not yet supported by any structuredschema, and to answer old questions more accurately. For example, if we learn to accurately predict the surface form relation "X-is-scientist-at-Y", this can help us to better predict the Freebase employee(X,Y) relation. To populate a database of such schema we present a family of matrix factorization models that predict affinity between database tuples and relations. We show that this achieves substantially higher accuracy than the traditional classification approach. More importantly, by operating simultaneously on relations observed in text and in pre-existing structured DBs, we are able to reason about unstructured and structured data in mutually-supporting ways. By doing so our approach outperforms state-of-the-art distant supervision.
Rob Procter, University of Warwick
The Opportunities and Challenges of Big Social Data Analytics
Thursday 6th March 2014, 4pm, CS1.01
The recent explosion of social media in the form of blogs, micro-blogs, social networking platforms and other ‘born-digital’ social data presents significant opportunities and challenges to researchers. In this talk, I will present the Collaborative Online Social Media Observatory, a social data analytics platform and use examples of current research to illustrate how it may help to meet these challenges.
Rob is Professor of Social Informatics in the Department of Computer Science and Exchange Professor, NYU. One focus of his current work is methodologies and tools for big social data analytics. Rob led a multidisciplinary team working with the Guardian/LSE on the ‘Reading the Riots’ project, analysing tweets sent during the August 2011 riots. This won the Data Visualization and Storytelling category of the 2012 Data Journalism Awards and the 2012 Online Media Award for the ‘Best use of Social Media’. Rob is one of the core members of the Warwick Institute for the Science of Cities (WISC), the UK partner of the NYU-led Center for Urban Science and Progress. He is also a co-founder of the Collaborative Online Social Media Observatory (Cosmos), a multidisciplinary group of UK researchers building a platform for social data analytics.
Nigel Smart, University of Bristol
Computing on Encrypted Data
Thursday 20th February 2014, 4pm, CS1.01
In the last couple of years amazing advances have been made on techniques to perform computation on encrypted data. Some of the techniques are even becoming practical. In this talk I will show a novel technique which utilizes techniques used in Fully Homomorphic Encryption (FHE) schemes to provide efficiency improvements in Multi-Party Computation (MPC) protocols. No prior knowledge of FHE or MPC will be assumed.
Prof. Nigel Smart is a leading world expert in cryptography. He currently serves Vice-President of the IACR and has in the past been both Programme and General Chair of Eurocrypt. Before joining the University he worked at Hewlett-Packard's European research laboratory; which is also located in Bristol. Prof. Smart has contributed to cryptography in many areas including elliptic curves, pairing based cryptography, protocol design and analysis (particularly key agreement), and more recent in multi-party computation and fully homomorphic encryption. He is the holder of an ERC Advanced Grant (2011-2016) and was the holder of a Royal Society Wolfson Merit Award (2008-2013). Prof. Smart has regular contacts with industry, and has consulted for a number of companies. He was also one of the founders of Identum, a spin-out in the area of encryption from the University, which has since been bought by Trend Micro.
Amanda Clare, Aberystwyth University
Bioinformatics and computational biology: 500 years of exciting problems?
Thursday 16th January 2014, 4pm, CS1.01
In a 1993 interview, Donald Knuth worried that computer science in the future will be "pretty much working on refinements of well-explored things", whereas "Biology easily has 500 years of exciting problems to work on". I'll describe some of the bioinformatics and computational biology that I've been working on. I'll also talk a little about where the field has come from, where it's going in the future, and whether it should be considered a branch of computer science at all.
Luca Cardelli, Microsoft Research
The Cell Cycle Switch Computes Approximate Majority
Thursday 9th January 2014, 4pm, CS1.01
Biological systems have been traditionally, and quite successfully, studied as 'dynamical systems', that is, as continuous systems defined by differential equations and investigated by sophisticated analysis of their continuous state space. More recently, in order to cope with the combinatorial complexity of some of these systems, they have been modeled as 'reactive systems', that is, as discrete systems defined by their patterns of interactions and investigated by techniques that come from software and hardware analysis. There are growing formal connections being developed between those approaches, and tools and techniques that span both. The two approaches can be usefully combined to bring new insights to specific systems. In one direction we can ask 'what algorithm does a dynamical system implement' and in the opposite direction we can ask 'what is the dynamics of a reactive system as a whole'. Answers to these questions can establish links between the structure of a system, which is dictated by the algorithm it implements, and the function of the system, which is represented by its dynamic behavior. Since there is depth on both sides, in the intricacies of the algorithms, and in the complexity of the dynamics, a better understanding can emerge of whole systems. I will focus in particular on a connection between a clever and well-studied distributed computing algorithm, and a simple chemical system (4 reactions). It leads to a connection between that algorithm and a well-known biological switch that is universally found as part of cell cycle oscillators. I will also discuss a general network structure for oscillators, based on the above switches, and how they are implemented 'in practice' in natural systems. These connections are examples of 'network transformations' that preserve both structure and functionality across systems of different complexity. Joint work with Attila Csikász-Nagy.
Boris Motik, University of Oxford
Theory and Practice: The Yin and Yang of Intelligent Information Systems
Thursday, 5th December 2013, 4pm, CS1.01
Theoretical and practical research complement each other, and their synergy is essential to developing advanced information systems. I will illustrate this using examples from his research in ontology languages, reasoning, and big data management. I will also discuss how we can foster interest in combining theory and practice in undergraduate and postgraduate education.
Jeff Yan, University of Newcastle
Why is usable security hard?
Thursday, 28th November 2013, 4pm, CS1.01
A major development in computer security in the past decade is the emergence of "usable security", which has now become a thriving and fast-moving discipline. Most, if not all, people agree that we need security systems that are simultaneously secure and usable. However, it is hard to design such solutions. I will illustrate the challenges of achieving usable security by examining the recent development of graphical password research, which aims to deliver a graphical alternative to the ubiquitous textual password scheme (that has long suffered from usability problems). Some novel attacks and open problems will be highlighted. To make better graphical passwords, expertises in security, usability and signal processing all matter.
Andreas Vlachos, University of Cambridge
The SpaceBook project: Assisting tourists in navigating and exploring a city
Thursday, 21st November 2013, 4pm, CS1.01
In this talk I will give an overview of the EU project SpaceBook, whose aim is to develop a speech-driven, hands-free, eyes-free tourism assistant. Then I will focus on two components of the system that rely on natural language processing, database population and user utterance interpretation. I will discuss the approaches and the resources developed for both tasks in detail and present evaluation results on real world data sets.
Stephen Pulman, University of Oxford
Thursday, 14th November 2013, 4pm, CS1.01
Sentiment Analysis - recognising positive and negative attitudes expressed in text - has become a very popular application of computational linguistics techniques, spawning a large number of startups, and generating a lot of commercial interest. In this talk I summarise recent research and trends in sentiment analysis, look at some relatively novel applications, and also critically examine various claims that have been made about the role of sentiment analysis in tasks like stock market prediction and election result forecasting.
Alina Ene, Princeton University/University of Warwick
Routing in Directed Graphs with Symmetric Demands
Thursday, 10th October 2013, 4pm, CS1.01
In this talk, we consider some fundamental maximum throughput routing problems in directed graphs. In this setting, we are given a capacitated directed graph. We are also given source-destination pairs of nodes (s_1, t_1), (s_2, t_2), …, (s_k, t_k). The goal is to select a largest subset of the pairs that are simultaneously routable subject to the capacities; a set of pairs is routable if there is a multicommodity flow for the pairs satisfying certain constraints that vary from problem to problem (e.g., integrality, unsplittability, edge or node capacities). Two well-studied optimization problems in this context are the Maximum Edge Disjoint Paths (MEDP) and the All-or-Nothing Flow (ANF) problem. In MEDP, a set of pairs is routable if the pairs can be connected using edge-disjoint paths. In ANF, a set of pairs is routable if there is a feasible multicommodity flow that fractionally routes one unit of flow from s_i to t_i for each routed pair (s_i, t_i).
MEDP and ANF are both NP-hard and their approximability has attracted substantial attention over the years. Over the last decade, several breakthrough results on both upper bounds and lower bounds have led to a much better understanding of these problems. At a high level, one can summarize this progress as follows. MEDP and ANF admit poly-logarithmic approximations in undirected graphs if one allows constant congestion, i.e., the routing violates the capacities by a constant factor. Moreover, these problems are hard to approximate within a poly-logarithmic factor in undirected graphs even if one allows constant congestion. In sharp contrast, both problems are hard to approximate to within a polynomial factor in directed graphs even if a constant congestion is allowed and the graph is acyclic.
In this talk, we focus on routing problems in directed graphs in the setting in which the demand pairs are symmetric: the input pairs are unordered and a pair s_i t_i is routed only if both the ordered pairs (s_i,t_i) and (t_i,s_i) are routed. Perhaps surprisingly, the symmetric setting can be much more tractable than the asymmetric setting. As we will see in this talk, when the demand pairs are symmetric, ANF admits a poly-logarithmic approximation with constant congestion. We will also touch upon some open questions related to MEDP in directed graphs with symmetric pairs.
This talk is based on joint work with Chandra Chekuri (UIUC).
Florin Ciucu, University of Warwick
Unification Day: On the Use of Martingales to Queueing and Caching Analysis
Thursday, 3rd October 2013, 4pm, CS1.01
This talk focuses on martingales, as very convenient stochastic processes to represent and handle information for two classical performance analysis problems: queueing and caching. We first present a martingale representation of a queueing system enabling a very simple handling of typical queueing operations: 1) multiplexing results in the multiplication of martingales and 2) scheduling results in time-shifting of the underlying martingales. Second we present a fractional change of measure technique, relying on a Radon-Nikodym density martingale, to analyze lines of time-to-live (TTL) caches. The elegance of the two martingale representations lies in the unified manner to address general classes of arrival patterns and caching policies, respectively.
Marian Gheorghe, University of Sheffield
Membrane computing - theory and applications
Thursday, May 23rd 2013. 4pm, CS1.01
Membrane computing has been introduced 15 years ago as a nature-inspired computational model generalising models, like Lindenmayer systems and DNA computing. The field has grown since then by considering different concepts related to cellular biology like cell structure, inner and trans-membrane operations, or different types of bio-chemical elements. Although originally rooted in formal language theory, the research in this field has soon started looking at connections with other computational models, like process algebras, Petri nets and cellular automata. This talk will briefly overview the key research topics in membrane computing, presenting some of the most investigated theoretical problems, including computability and complexity aspects. Some applications will be discussed and connections between membrane computing and formal verification based on model-checking will be pointed out.
Peter Ward, IBM
IBM Global Technology Outlook 2013
Thursday, May 16th 2013. 4pm, CS1.01
The Global Technology Outlook (GTO) is IBM Research’s vision of the future for information technology (IT) and its implications on industries. This annual exercise highlights emerging software, hardware, and services technology trends that are expected to significantly impact the IT sector in the next 3-10 years. In particular, the GTO identifies technologies that may be disruptive to an existing business, have the potential to create new opportunity, and can provide new business value to our customers. The 2013 GTO builds upon 100 years of IBM innovation as we embark on the era of context-centric cognitive computing. The confluence of mobile, social, cloud and analytics are enabling the creation, manipulation and exploitation of context. How much value is contained in growing volumes of images and videos? How can flexible interactive visualization technologies enhance business and scientific analytics and discovery? How can personalized education experiences be delivered through knowledge management, analytics and context? What mobile strategy will enable a secure end-to-end platform and solutions for constantly connected employees and customers? How are companies using the programmable web to externalize core offerings and enterprise capabilities to the global network? What benefits can be realized
from a federated heterogeneous cloud for just-in-time workload optimization? The 2013 GTO focuses on six topics: Contextual Enterprise, Future of Analytics: Multimedia and Visual Analytics, Mobile First, Scalable Services Ecosystem, Software Defined Environments, and the Future of Education.
Maria Liakata, University of Warwick
Text mining for knowledge discovery: domain theories, scientific discourse, emotions
Thursday, May 9th 2013. 4pm, CS1.01
The boom in the life sciences has fuelled interest into methods for automatic and high throughput processing of experimental data and scientific publications alike. Meanwhile the ever increasing numbers of documents of all kind in electronic form opens new opportunities for knowledge mining and discovery. I will be presenting work of mine along three different strands of knowledge discovery from textual data:
- the induction of domain theories (facts that describe a domain of interest) from newswire text
- work on scientific discourse annotation in life-science articles, which allows the characterisation of knowledge statements in the scientific literature, and how this is relevant to a range of e-Science applications, from the generation of alternative abstracts to updating the information in drug package descriptions and pathway curation
- work in recognising emotions from suicide notes and possible extensions.
Mark Nixon, University of Southampton
Gait and soft biometrics
Thursday, April 25th 2013. 4pm, CS1.01
The talk will discuss the current state-of-art in gait and soft biometrics. Gait is well established now for recognising people by the way they walk and I shall discuss how that has been achieved and the other areas of interest when recognising people by their walking - or running - pattern. Soft biometrics is an emerging area of interest in biometrics: can we augment computer vision derived measures by human descriptions and if so, what is the interrelationship between them. We have been developing new approaches in gait biometrics, based on human descriptions which can extend the descriptions derived by automatic means, say by using computer vision. The human descriptions are semantic and are a set of labels which are converted into numbers. Naturally, there are considerations of language and psychology when the labels are collected. This talk will describe how the labels are collected, how they can be used to enhance recognising people by the way they walk (we also use the approaches to recognise vehicles by sound). A new development of this approach might lead to a new procedure for collecting witness statements.
Mark Nixon is the Professor in Computer Vision at the University of Southampton UK. His research interests are in image processing and computer vision. His team develops new techniques for static and moving shape extraction which have found application in automatic face and automatic gait recognition and in medical image analysis. His team were early workers in face recognition, later came to pioneer gait recognition and more recently joined the pioneers of ear biometrics. Amongst research contracts, he was Principal Investigator with John Carter on the DARPA supported project Automatic Gait Recognition for Human ID at a Distance and he was previously
with the FP7 Scovis project and is currently with the EU-funded Tabula Rasa project.
Mark has published over 400 papers in peer reviewed journals, conference proceedings and technical books His vision textbook, with Alberto Aguado, Feature Extraction and Image Processing (Academic Press) reached 3rd Edition in 2012 and has become a standard text in computer vision. With Tieniu Tan and Rama Chellappa, their book Human ID based on Gait is part of the Springer Series on Biometrics and was published in 2005. He has chaired/program chaired many conferences (BMVC 98, AVBPA 03, IEEE Face and Gesture FG06, ICPR 04, ICB 09, IEEE BTAS 2010) and given many invited talks. Mark is currently chair of the IAPR Fellow Committee and is a member of IAPR TC4 Biometrics and of the IEEE Biometrics Council. Dr. Nixon is a Fellow IET and FIAPR.
Andrzej Murawski, University of Warwick
Programming with higher-order storage
Thursday, March 7th 2013. 1pm, CS1.01
We shall consider a higher-order programming language that allows the programmer to store arbitrary functions in variables. I will discuss the computational power of this paradigm and present some recent expressivity results. In particular, it turns out that any uses of storable functions can be faithfully simulated with a single memory cell for storing functions of type unit->unit.
This is joint work with Nikos Tzevelekos (Queen Mary, University of London), to appear in FOSSACS later this year.
Tim Furche and Christian Schallhart, University of Oxford
DIADEM: Domains to Databases
Thursday, January 17th 2013. 4pm, CS1.01
What if you could turn all websites of an entire domain into a single database? Imagine all real estate offers, all airline ﬂights, or all your local restaurants’ menus automatically collected from hundreds or thousands of agencies, travel agencies, or restaurants, presented as a single homogeneous dataset. Historically, this has required tremendous effort by the data providers and whoever is collecting the data: Vertical search engines aggregate offers through speciﬁc interfaces which provide suitably structured data. The semantic web vision replaces the speciﬁc interfaces with a single one, but still requires providers to publish structured data. Attempts to turn human-oriented HTML interfaces back into their underlying databases have largely failed due to the variability of web sources.
In this talk, we demonstrate that this is about to change: The availability of comprehensive entity recognition together with advances in ontology reasoning have made possible a new generation of knowledge driven, domain-speciﬁc data extraction approaches. To that end, we give an overview of DIADEM, the ﬁrst automated data extraction system that can turn nearly any website of a domain into structured data, working fully automatically, and present some preliminary evaluation results (see also
http://diadem.cs.ox.ac.uk/ [diadem.cs.ox.ac.uk] ). This talk describes joint work with Georg Gottlob, Giovanni Grasso, Giorgio Orsi, as well as with all other members of the DIADEM Team.
Matthias Vigelius, Monash University
GPU-accelerated stochastic simulations of collective behaviour
Wednesday, December 12th 2012. 2pm, CS1.04
Reaction-diffusion systems can be used to model a large variety of complex self-organized phenomena occurring in biological, chemical, and social systems. Such models are usually described on the macroscopic level with a Fokker-Planck equation. Examples from the literature where this approach has successfully been applied include cell migration, social insect foraging, quorum sensing in bacteria, and invasion processes. However, a macroscopic approach means that it is impossible to incorporate experimental observations on the individual level or hypotheses about the individual behaviour in a principled way.
On the other hand, Langevin-type individual-based microscopic models can in principle address these issues. However, a statistically faithful implementation of complex Langevin models is challenging, and as yet no implementation exists that incorporates inhomogeneous drift fields and inhomogeneous diffusivity. Since such simulations are also computationally expensive, hardware limitations severely restrict their applicability to models of realistic size. Thus, to analyse individual-based Langevin models
of systems that would normally be modelled on the aggregate level, the performance of the corresponding algorithms is inadequate and speed-ups of several orders of magnitude are required. Parallelization is a natural approach to attain the desired performance increase and graphics processors provide a cheap and accessible hardware platform for high performance implementations of massively parallel algorithms.
In this talk, I will present two methods that we assessed in order to approach this problems. I will talk about their respective pros and cons and how they lend themselves to parallelization on graphics hardware. I will present Inchman, an easy-to-use web framework that allows researchers to set up and run biological simulations on GPU clusters. I will also give some examples on applications from various fields where we have successfully employed Inchman.
The speaker is a Research Fellow at Monash FIT, Centre for research in intelligent systems (CRIS), PhD University of Melbourne 2009.
Victor Sanchez, University of Warwick
Security in mobile ad hoc networks: attacks and countermeasures
Thursday, November 29th 2012. 4pm, CS1.01
A mobile ad hoc network (MANET) is a collection of mobile nodes equipped with wireless communication capabilities. These nodes dynamically form a temporary network without the need of any existing network infrastructure. The absence of the central infrastructure, in conjunction with an open peer-to-peer network architecture, frequent changes of the topology and a shared wireless medium, pose important challenges to security design. This seminar presents a review of the main security problems of MANET and the state-of-the-art security proposals.
Daniel Kral, University of Warwick
Testing first order properties in sparse combinatorial structures
Thursday, November 15th 2012. 4pm, CS1.01
Algorithmic metatheorems guarantee that certain types of problems have efficient algorithms. A classical example is the theorem of Courcelle asserting that every MSO
property can be tested in linear time for relational structures with bounded tree-width. As examples of MSO properties of graphs, let us mention 3-colorability and hamiltonicity, both well-known NP-hard problems.
In this talk, we focus on simpler properties, those that can be expressed in first order logic (FOL). An example of FO property is an existence of a fixed substructure. While it is not hard to show that every FO property can be decided in polynomial time, our desire is to design algorithms with faster running time (e.g. linear time). We recall a recent notion of graph classes with bounded expansion, which include classes of graphs with bounded maximum degree and proper-minor closed classes of graphs. We then apply structural results to show that FO properties can be tested in linear time for classes of graphs with bounded expansion and we will discuss extensions to other structures. At the end of the talk, we will mention several open problems as well as directions for future research.
The talk is based on joint results with Zdenek Dvorak (Charles University, Prague) and Robin Thomas (Georgia Institute of Technology, Atlanta).
Rajeev Raman, University of Leicester
Range Extremum Queries
Thursday, November 8th 2012. 4pm, CS1.01
There has been a renewal of interest in data structures for range extremum queries. In such problems, the input comprises N points. These points are either elements of a d-dimensional matrix, that is, their coordinates are specified by the 1D submatrices they lie in (row and column indices for d=2), or they are points in R^d. Furthermore, associated with each point is a priority that is independent of the point's coordinate. The objective is to pre-process the given points and priorities to answer the following quer(ies): given a d-dimensional rectangle, report the points with maximum (or minimum) priority (range max query) or, for some fixed parameter k, report the points with the k largest (smallest) priorities. The objective is to minimize the space used by the data structure and the time taken to answer the above query. This talk surveys a number of recent developments in this area, focussing on the cases d=1 and d=2.
After defending his PhD thesis in October 1991, Rajeev took up a Postdoctoral Fellowship in the Algorithms and Complexity Group at the Max-Planck-Institut für Informatik, which is headed by Kurt Mehlhorn. In January 1993 he joined the University of Maryland Institute for Advanced Computer Studies, as a Research Associate working with Uzi Vishkin. Crossing the Atlantic yet again, Rajeev joined the Algorithm Design Group at King's College London in 1994. He has been at Leicester since January 2001.
Prof. Raman has also been a visiting researcher at the Max-Planck-Institut and at Hong Kong UST. He taught high school students at the Johns Hopkins Center for Talented Youth.
Francois Taiani, Lancaster University
Geology: Modular Georecommendation In Gossip-Based Social Networks
Thursday, October 4th 2012. 4pm, CS1.01
Geolocated social networks, that combine traditional social networking features with geolocation information, have grown tremendously over the last few years. Yet, very few works have looked at implementing geolocated social networks in a fully distributed manner, a promising avenue to handle the growing scalability challenges of these systems. In this talk, I will focus on georecommendation, and show that existing decentralized recommendation mechanisms perform in fact poorly on geodata. I will present a set of novel gossip-based mechanisms to address this problem, which we have captured in a modular similarity framework called Geology. The resulting platform is lightweight, efficient, and scalable, and I will illustrate its superiority in terms of recommendation quality and communication overhead on a real data set of 15,694 users from Foursquare, a leading geolocated social network.
This is joint work with Anne-Marie Kermarrec (INRIA Rennes, France) and Juan M. Tirado (Universidad Carlos III, Spain). The talk is based on our joint publication at ICDCS 2012.
Francois has been a lecturer at Lancaster since January 2005, after an intervening spell as a post-doctoral researcher at AT&T Shannon Laboratory (NJ, USA), on an INRIA scholarship. He received his PhD in January 2004 for his work at LAAS-CNRS (France) on multi-level reflection applied to fault-tolerant systems. At LAAS, Francois has worked among others on the EU ISP Dependable Systems of Systems project (IST-1999-11595), and the EU Cabernet Network of Excellence. At Lancaster he is co-investigator on the Divergent Grid project (EPSRC EP/C534891), and the EU FP7 Diva project (Dynamic Variability in complex, Adaptive systems).
Iain Stewart, University of Durham
Some recent developments in interconnection networks
Thursday, May 10th 2012. 4pm, CS1.01
Interconnection networks play a fundamental role in Computer Science. They are the means by which the processors of a distributed-memory multiprocessor computer communicate and also by which I/O devices communicate with processors and memory; they are increasingly used in network switches and routers to replace buses and crossbars; and they are crucial to on-chip networks. Their usage is becoming more prevalent and on an increasingly grand scale, as we manage to incorporate more and more 'nodes' within the confines of a single space. For example, the newest generation of high-end supercomputer networks is beginning to use high(er) dimensional tori: the IBM Blue Gene/Q is based around a 5-dimensional torus; and the K computer is based around a 6-dimensional torus. The extended applications of interconnection networks allied with technological advances promotes (the currently theoretical) investigation of new interconnection networks with a view to ensuring properties beneficial to the domain of application. In this talk, we'll look at some basic topological issues relating to (some popular) interconnection networks before moving on to look at relatively more recent hierarchical interconnection networks, some of which have been inspired by the use of free-space optical interconnect technologies in combination with electronic technologies. The talk will be suitable for a general audience and theoretical in nature.
Iain is a Professor of Computer Science within the School of Engineering and Computing Sciences at Durham. He read mathematics at Christ Church, Oxford before completing his PhD in mathematics at Queen Mary College, University of London. He was then appointed to a Lecturership in the Computing Laboratory, University of Newcastle upon Tyne in 1986 before moving to a Lecturership in the Department of Computer Science, University of Wales Swansea in 1992. He was later appointed Senior Lecturer and then Reader before becoming Professor of Computer Science at Leicester in 1996. He moved to Durham in 2002. He has had number of positions within the Computer Science community, some of which have been under the auspices of the London Mathematical Society which has provided strong support for Computer Science for quite a few years.
Mathai Joseph, Tata Consultancy Services
The End of 101, 202?
Friday, May 4th 2012. 2pm, CS1.01
For the last few decades, university education in many countries has faced a number of conflicting pressures: on one hand to to attract the best students and provide high quality education and on the other to increase class sizes and reduce the cost of education. It now faces a new challenge as online courses begin to offer ways of changing undergraduate education even further. There is uncertainty about how undergraduate education will be affected and in some cases universities have responded by deciding to offer their own online education.
These are early days for online education and its supporters and funders are trying to guess where it will be successful. It is clear that the technology is still evolving and there are a number of areas where a great deal more work is needed before online education can provide a learning experience that matches more traditional undergraduate education. Nevertheless, online education may well be the only solution possible in countries like India where the demand for university education has far outstripped the ability of universities to provide it.
Mathai Joseph was at the Tata Institute of Fundamental Research in Mumbai until 1985 when he became a professor of computer science at the University of Warwick. In 1997 he returned to India and became Executive Director at the Tata Research Development and Design Centre, the R&D division of Tata Consultancy Services. He is now an advisor to TCS.
Graham Cormode, AT&T Labs Research
Thursday, March 15th 2012. 4pm, CS1.01
In dealing with massive distributed data, exact computations may be possible but can still be very expensive to perform. Instead, it is often desirable to look to more lightweight computations that give (guaranteed) approximations. A central approach is the idea of the mergeable summary: a compact data structure that summarizes a data set, which can be merged with others to summarize the union of the data without significant growth in size. This framework enables parallel computation.
Samples and sketches are two examples of well-known, effective mergeable summaries. I'll present recent results on new, compact mergeable summaries for various tasks, such as approximating frequent items, order statistics, and range counts.
Joint work with Pankaj Agarwal, Zengfeng Huang, Jeff Phillips, Zhewei Wei, Ke Yi
I am a researcher at AT&T Labs--Research in New Jersey. I work on data stream analysis, massive data sets, and general algorithmic problems.
Before this, I worked at Lucent Bell Laboratories, with focus on Network Management, and previously, I was a Postdoc researcher at the DIMACS research facility, which is located at Rutgers University. I retain connections to DIMACS and to MassDAL -- the Massive Data Analysis Lab. I did my PhD work at the Department of Computer Science at the University of Warwick, UK, and spent some time studying in Cleveland, Ohio at Case Western Reserve University with the Electrical Engineering and Computer Science Department .
Ian Stark, University of Edinburgh
Exploring variation in biochemical pathways with the continuous pi-calculus
Thursday, February 23rd 2012. 4pm, CS1.01
Theoretical computer science and biology have common cause in trying to model and understand complex interacting processes. One potential contribution from computer science is the use of high-level languages for concurrent systems to smooth the route between natural descriptions and mathematically precise models. This is similar to the relation between programming languages and compiled executables: these languages allow us to express and communicate intuition, while keeping contact with low-level realities.
The continuous pi-calculus is one such language for modelling biochemical systems. It is based on Milner's pi-calculus, a language of interacting processes with dynamic communication topology. Continuous pi modifies this to express the real-valued concentrations, reaction rates, and multiway interactions of biochemistry. Models in continuous pi are reagent-centred descriptions of individual chemical species, expressing their potential behaviour and interactions. These descriptions can be compiled into ordinary differential equations; and numerical simulation then shows the behaviour over time of biochemical mixtures: reactions, complexes, enzymes, activation and inhibition.
As well as modelling specific biochemical systems, these high-level descriptions allow us to investigate potential variations to those systems, and to explore the accessible evolutionary landscape. To demonstrate this we modelled a classic intracellular signalling pathway and then simulated every variant system reachable by a single change. We then used assertions in temporal logic to evaluate how well these variant pathways carried a signal. This complete survey of the one-step neighbourhood highlights which parts of the pathway are robust, maintaining behaviour under change, and which parts offer potentially new behaviours.
This investigation of system behaviour under variation gives an example of how high-level languages for describing biological processes make it possible to express and test correspondingly high-level hypotheses about robustness, neutrality and evolvability.
Work on continuous pi is a collaboration with Marek Kwiatkowski and Chris Banks.
Ian Stark is a Senior Lecturer in Computer Science at the University of Edinburgh School of Informatics, where he works in the Laboratory for Foundations of Computer Science. His research is on mathematical models for programming languages and concurrent interacting systems, in particular reasoning about names, local state, and mobility. His work on biochemical modelling is in association with the SynthSys Centre for Integrative Systems Biology at Edinburgh (formerly CSBE).
Samin Ishtiaq, Microsoft Research
Separation Logic and Device Drivers
Thursday, February 16th 2012. 4pm, CS1.01
Device drivers are a major cause of OS crashes and hangs. As drivers spend much of their time handling queues of requests, and these requests are stored as mutable linked lists, many of these errors come from not maintaining a memory safety invariant. SLAyer is a program analysis tool designed to prove the absence of memory safety errors such as dangling pointer references and double frees. Towards this goal, SLAyer searches for invariants that form proofs in Separation Logic. Complex composite data structures like cyclic doubly linked-lists are supported using parametrized recursive predicates. SLAyer is aimed at moderately sized (10-30K) systems-level code bases written in C; it is fully automatic and does not require annotations or hints from the programmer.
Samin Ishtiaq is Principal RSDE in the Programming Principles and Tools group at Microsoft Research Cambridge. He works on the SLAyer (Separation Logic-based memory safety for C programs), TERMINATOR (program termination) and BioCheck (analysis of gene regulatory networks) projects. Samin joined MSR in April 2008. Before that, he worked at ARM, where he did CPU modeling and verification to help tape-out the Cortex A8, Cortex M3 and SC300 processors, and the AMBA bus protocol checker. Samin has an MEng from Imperial and a PhD in dependent type theory from Queen Mary. He was an RA on the original Verified Bytecode project.
Nick Jennings, Southampton University
Computational Service Economies: Design and Applications
Thursday, January 26th 2012. 4pm, CS1.01
Many modern computing systems have to operate in environments that are highly interconnected, highly unpredictable, without a central control authority, and in which the constituent components are owned by a variety of stakeholders that each have their own aims and objectives. Relevant exemplars include the Web, the Smart Electricity Grid, Peer-to-Peer systems, Pervasive Computing and many eCommerce applications. Now, I believe that all of these systems can operate under the same fundamental conceptual model: (i) entities offer a variety of services in some form of institutional setting; (ii) other entities connect to these services (covering issues such as service discovery, service composition and service procurement); and (iii) entities enact services in a flexible and context sensitive manner. Moreover, I believe agent-based computing is an appropriate computational model for such systems. In particular, autonomous agents are a natural way of viewing flexible service providers and consumers and the interactions between these autonomous components are naturally modeled as some form of economic trading process that, if successful, results in a transaction between the agents involved.
In this talk, the focus will be on the methods and techniques for designing the electronic institutions and the interaction strategies of the participants. In so doing, I will touch upon techniques from the areas of game theory, coalition formation, automated negotiation and computational mechanism design. Relevant exemplars of applications built using such techniques in the domains of sensor networks, autonomous systems, smart grids, and disaster response will also be discussed.
Professor Jennings divides his time between his posts as a Chief Scientific Advisor to the UK Government and Professor of Computer Science in the School of Electronics and Computer Science at Southampton University, where he heads the Agents, Interaction and Complexity Group (he previously headed the Intelligence, Agents, Multimedia Group). He is also the Chief Scientist for aroxo and lostwax/aerogility.
Nick is an internationally-recognised authority in the areas of agent-based computing and intelligent systems. His research covers both the science and the engineering of such systems. Specifically, he has undertaken fundamental research on automated bargaining, auctions, markets, mechanism design, trust and reputation, coalition formation and decentralised control. He has also pioneered the application of multi-agent technology; developing some of the first real-world systems (in domains such as business process management, energy systems, sensor networks, disaster response, telecommunications, and eDefence) and generally advocating the area of agent-oriented software engineering.
In undertaking this research, he has attracted grant income of over £20M (mainly from EPSRC), published more than 500 articles (with some 250 co-authors) and graduated 30 PhD students (two of whom have won the BCS/CPHC Distinguished Dissertation Award). He is recognised as highly cited by ISI Web of Science in both the Engineering and the Computer Science categories. With over 40,000 citations in Google Scholar, he is the second most highly cited researcher in the area of artificial intelligence (according to Microsoft's Academic Search system) and has an h-index of 86 (the second top non-American according to Palsberg). He has received a number of international awards for his research: the Computers and Thought Award (the premier award for a young AI scientist and the first European-based recipient in the Award's 30 year history), the ACM Autonomous Agents Research Award and an IEE Achievement Medal. He is a Fellow of the Royal Academy of Engineering, the Institute of Electrical and Electronic Engineers, the British Computer Society, the Institution of Engineering and Technology (formerly the IEE), the Association for the Advancement of Artificial Intelligence (AAAI), the Society for the Study of Artificial Intelligence and Simulation of Behaviour (AISB), and the European Artificial Intelligence Association (ECCAI) and a member of Academia Europaea and the UK Computing Research Committee (UKCRC).
Nick was the founding Editor-in-Chief of the International Journal of Autonomous Agents and Multi-Agent Systems, is a member of the scientific advisory board of the German AI Institute (DFKI) and a founding director of the International Foundation for Autonomous Agents and Multi-Agent Systems. He has also led teams that have won competitions in the areas of: the Iterated Prisoners' Dilemma (the 20th Anniversary competitions in 2004 and 2005), RoboCup Rescue (the Infrastructure competition in 2007), Agent Trust and Reputation (the ART competitions in 2006 and 2007), the Lemonade Stand Game (2009 & 2010), and Market Design (the TAC CAT competition in 2007).
Martin Lippert, VMware
The daily software engineering life - How to be prepared
Thursday, January 12th 2012. 4pm, CS1.01
How is life as a real software engineer, out there in the wild? This talk gives an overview of today's software engineering in practice. We talk about what it means to build and ship software in various kinds of companies, ranging from modern and flexible ones like Google or Facebook to more traditional ones like large insurance companies, for example. We discuss the most important challenges people are facing when implementing, testing, shipping and supporting software systems, the role of soft-skills in this area, why you need to understand the domain you are building software for, and what it means to ship new versions of a software several times a day. We talk about the role of agile software development methods and how they changed the landscape, why they are an essential part of today's software engineering - and what people should better have learned at University before jumping onto this playground.
Martin works at VMware, where he leads the tooling development team - the team that is responsible for the Spring IDE, the SpringSource Tool Suite and the Cloud Foundry Integration for Eclipse. His work is focused on building development tools, working with the community, and creating a flexible, modern and agile development process with his team. Before joining VMware Martin founded it-agile GmbH, one of the leading agile consulting companies in Germany, where he worked with teams across Germany and Europe on agile software development, flexible and modular architectures and Eclipse technologies. He is committer for various open-source projects, author of a number of articles about agile software development, and co-author of several books. He is also a frequent speaker at non-academc software engineering conferences. Martin got a Diploma in Informatics from the University of Hamburg in 1999, with a focus on software engineering and object-orientation.
Richard Dearden, University of Birmingham
Model-based fault diagnosis for an autonomous underwater vehicle
Thursday, December 1st 2011. 4pm, CS1.01
In this talk I describe work on the automatic fault diagnosis of an autonomous underwater vehicle (AUV), using a model-based approach. I will describe Livingstone 2, a discrete fault diagnosis system that has been deployed in a number of applications including spacecraft, and how this was used to diagnose Autosub 6000, a research AUV operated by the National Oceanography Centre. I will also look at some of the difficulties encountered in applying such a discrete approach, and describe recent research on hybrid diagnosis for the domain using algorithms built around a satisfiability modulo theory solver.
Dr. Richard Dearden has been a Senior Lecturer in the School of Computer Science at the University of Birmingham since 2005. He works in the broad area of reasoning under uncertainty, and in particular on planning, execution, and fault diagnosis, all applied to autonomous robots. Before that he spent 5 years leading the Model-based Diagnosis and Recovery group at NASA Ames Research Centre where he worked primarily on Mars rover diagnosis and planning. His Ph.D. (2000) was in Markov decision process planning at the University of British Columbia.
Jose Fiadeiro, University of Leicester
A Formal Approach to Service-Oriented Modelling
Thursday, November 24th 2011. 4pm, CS1.01
This talk provides an overview of a formal approach that we have developed within the FP6-IST-FET Integrated Project SENSORIA (www.sensoria-ist.eu [www.sensoria-ist.eu]), which aimed at providing formal support for modelling service-oriented systems in a way that is independent of the languages in which services are programmed and the platforms over which they run. We discuss the semantic primitives that are provided in the SENSORIA Reference Modelling Language (SRML) for modelling composite services, ie. services who business logic involves a number of interactions among more elementary service components as well the invocation of services provided by external parties. This includes a logic for specifying stateful, conversational interactions, a language and semantic model for the orchestration of such interactions, and an algebraic framework supporting service discovery, selection and dynamic assembly.
I did my undergraduate degree in Mathematics at the University of Lisbon (Faculty of Science), after which I moved to the Technical University of Lisbon (Department of Mathematics, Faculty of Engineering) where I studied for a PhD under the supervision of Amilcar Sernadas. I was awarded my doctorate in 1989, and then spent three years doing research at Imperial College London with a grant from the European Commission. I became Associate Professor in Computer Science at the Technical University of Lisbon in 1992, and moved to the University of Lisbon (Department of Informatics, Faculty of Science) in 1993. Before I joined Leicester in 2002, I held visiting research positions at Imperial College, King’s College London, PUC-Rio de Janeiro, and the SRI International. I was Head of Department at Leicester between August 2006 and July 2011.
My current research interests are in formal aspects of software system modelling and analysis in the context of global ubiquitous computing. I am a member of the Steering Committees of WS-FM (Workshop on Web Services and Formal Methods), CALCO (Conference on Algebra and Coalgebra in Computer Science, which I co-founded with Jan Rutten) and WADT (Workshop on Algebraic Development Techniques). I was chairman of the IFIP WG 1.3 (Foundations of System Specification) in 2004-09, and chairman of the Steering Committee of ETAPS (European Joint Conferences on Theory and Practice) in 2002-04. I am also member of the Editorial Board of Information Processing Letters (Elsevier).
Jon Timmis, University of York
From Immune Systems to Robots
Thursday, November 10th 2011. 4pm, CS1.01
There are many areas of bio-inspired computing, where inspiration is taken from a biological system to construct an engineered solution. This talk will focus on the use of the immune system and its application to anomaly detection of chemical spectra in a robotic sniffer dog and to self-healing swarm robotic systems. In the first case, we will explore work undertaken with Dstl where we are attempting to develop an automated improvised explosive device (IED) detection system. For this work, we have abstracted signalling mechanisms from T cells, an important cell in an immune response. In the second study, one of self-healing swarm systems, we explore how we have exploited an immune response known as granuloma formation, to create a swarm that is capable to deciding how best to cope with energy failure scenarios. We will also explore the pitfalls of a bio-inspired approach and illustrate how computational modelling of immune systems can aid in bio-inspird algorithm design.
Jon Timmis is Professor of Natural Computation at the University of York and holds a joint appointment between the Department of Electronics and the Department of Computer Science. His primary research is in the area of computational immunology and bio-inspired fault tolerance in embedded systems, with a focus on swarm robotic systems. He gained his PhD in Computer Science from the University of Wales, Aberystwyth. He holds a senior Royal Society Research fellowship, a Wolfson Research Merit Award, to investigate the development of self-healing swarm robotic systems.
Jean-Christophe Olivo-Marin, Institut Pasteur
Cells, images and numbers
Thursday, October 20th 2011. 4pm, CS1.01
I will present specific methods and algorithms for the processing and quantification of 2- and 3-D+t images sequences in biological microscopy and demonstrate algorithms of PSF approximations for image deconvolution, image segmentation, multi-particle tracking and active contours models for cell shape and deformation analysis. One specific goal in biological imaging is indeed to automate the quantification of dynamics parameters or the characterization of phenotypic and morphological changes occurring as a consequence of cell/cell or pathogens/host cells interactions. The availability of this information and its thorough analysis is of key importance to help deciphering the underlying molecular mechanisms. I will illustrate our methods in projects related to the study of the dynamics of genes in cell nuclei, the movement of parasites in cells and the detection and tracking of microbes in cells.
J.-C. Olivo-Marin, PhD, is the head of the Quantitative Image Analysis Unit at the Institut Pasteur. He was a co-founder and Chief Technology Officer of the Institut Pasteur Korea, Seoul. He has a long experience of multi-disciplinary approaches in biological imaging, and his research interests are in image analysis of multidimensional microscopy images, pattern recognition and motion analysis for cellular dynamics studies. He is a member of the Bio Imaging and Signal Processing Technical Committee (BISP-TC) of the IEEE Signal Processing Society.
Ton Coolen, King's College London
Counting and generating tailored random graphs
Thursday, October 13th 2011. 4pm, CS1.01
Ensembles of tailored random graphs with controlled topological properties are a natural and rigorous language for describing biological networks. They suggest precise definitions of structural features, allow us to classify networks and obtain precise (dis)similarity measures, provide `null models' for hypothesis testing, and can serve as efficient proxies for real networks in process modelling. Mathematically and computationally, the key questions are (i) how to calculate the Shannon entropy of tailored random graphs analytically (giving the effective number of graphs with given imposed topological features) and (ii) how to generate such graphs numerically, with any specified probability distribution. Surprisingly, most algorithms used in the past for generating complex graphs are biased, rendering many studies on biological networks meaningless. We show how both problems can be solved exactly, using information-theoretic and statistical mechanical tools.
Coolen obtained his PhD in theoretical physics from the University of Utrecht; he came to the Oxford in 1991, and joined King's College London in 1995. He is a member of the Advisory Panel of Journal of Physics A, the BBSRC Systems Biology Board, and a Fellow of the London Institute for Mathematical Sciences. He has published two books, more than 130 refereed papers, and supervised 17 PhD students; for details see http://www.mth.kcl.ac.uk/~tcoolen. Coolen specialises in the mathematical analysis of large heterogeneous many-variable systems in physics, biology, computer science, and economics, statistical mechanics and stochastic processes, Bayesian and information-theoretic analysis, and the theory of complex networks. In recent years he has focused his research mainly on problems in bio-medicine. He created and led for many years the Disordered Systems group in Mathematics at King's College, and is now setting up an Institute for Mathematical and Molecular Biomedicine at King's, see http://www.mth.kcl.ac.uk/IMMB, which aims to become a leading centre for the development of advanced mathematical and computational methods for modern biomedicine.
Paulo Mateus, SQIG - Instituto de Telecomunicações Lisbon
Minimizing probabilistic and quantum automata
Monday, September 26th 2011. 2pm, CS1.01
Quantum gadgets are becoming available in the market, namely for communicating over optical fiber. The major engineering problem concerning such gadgets is the size of quantum memory that can be entangled. While it is more or less trivial to generate two entangled qubits, it is still science fiction to entangle much more than ten qubits. When working with such restrictions, a basic question occurs: what is the minimal number of (entangled) qubits required to perform a certain task? This question is equivalent to an open problem posed by Moore and Cruchfield in 2000 concerning the decidability of minimizing quantum automata. Furthermore, this question is already connected with another open problem from 1979, by Paz, concerning the existence of a Myhill-Nerode minimization algorithm for stochastic machines/probabilistic automata. We show that both these problems are decidable, and that there exists an EXPSPACE minimization algorithm for both quantum and probabilistic automata. Joint work with D. Qiu.
Born in 1975 in Lisbon, studied at IST where he got his PhD in 2001 with a thesis on the interconnection of probabilistic systems. In the Fall semester of 2001-02, he was a postdoc at the Logic and Computation Group, Department of Mathematics, University of Pennsylvania. In 2006, he obtained his agregação (habilitation) in Mathematics from the Technical University of Lisbon. His habilitation thesis was awarded the Portuguese IBM Scientific Prize 2005. Currently, he is an Associate Professor at the Department of Mathematics of IST and coordinates the Security and Quantum Information Group at Instituto de Telecomunicações.
Leslie Valiant, Harvard University
Tuesday, July 12th 2011. 4pm, MS.05
Using the notion of polynomial time reduction computer scientists have discovered an astonishingly rich web of interrelationships among the myriad computational problems that arise in diverse applications. These relationships can be used both to give evidence of intractability, such as that of NP-completeness, as well as to provide efficient algorithms.
In this talk we discuss a notion of a holographic reduction that is more general than the traditional one in the following sense. Instead of locally mapping solutions one-to-one, it maps them many-to-many but preserves some measure such as the sum of the solutions. One application is to finding new polynomial time algorithms where none was known before. Another is to give evidence of intractability. There are pairs of related problems that can be contrasted in this manner. For example, for a skeletal version of Cook's 3CNF problem (restricted to be planar and where every variable occurs twice positively) the problem of counting the solutions modulo 2 is NP-hard, but counting them modulo 7 is polynomial time computable. Holographic methods have proved useful in establishing dichotomy theorems, which offer a more systematic format for distinguishing the easy from the probably hard. Such theorems state that for certain wide classes of problems every member is either polynomial time computable, or complete in some class conjectured to contain intractable problems.
Leslie Valiant was educated at King's College, Cambridge; Imperial College, London; and at Warwick University where he received his Ph.D. in computer science in 1974. He is currently T. Jefferson Coolidge Professor of Computer Science and Applied Mathematics in the School of Engineering and Applied Sciences at Harvard University, where he has taught since 1982. Before coming to Harvard he had taught at Carnegie Mellon University, Leeds University, and the University of Edinburgh.
His work has ranged over several areas of theoretical computer science, particularly complexity theory, computational learning, and parallel computation. He also has interests in computational neuroscience, evolution and artificial intelligence.
He received the Nevanlinna Prize at the International Congress of Mathematicians in 1986, the Knuth Award in 1997, the European Association for Theoretical Computer Science EATCS Award in 2008, and the 2010 A. M. Turing Award. He is a Fellow of the Royal Society (London) and a member of the National Academy of Sciences (USA).
Mark Josephs, WMG Digital Laboratory
Processes through the Looking Glass: Reflections on an Algebra for Delay-Insensitive Circuits [slides]
Thursday, May 26th 2011. 4pm, CS1.04
In his 1988 Turing Award Lecture, Ivan Sutherland advocated the transition-signalling conceptual framework as it "offers the opportunity to build up complex VLSI systems by hierarchical composition from simpler pieces". The approach has its origins in the Macromodular Computer Design project at Washington University, St. Louis (1965-1974). In 1989, I started to collaborate with some Dutch researchers who had been developing its theoretical foundations and this collaboration quickly led to an algebraic approach (based on Hoare's Communicating Sequential Processes) to the design of delay-insensitive (DI) circuits. In this talk I shall give a short introduction to our so-called DI-Algebra, before focussing on more recent results based on the idea of constructing the mirror or reflection of a process.
Reference: M.B. Josephs and H.K. Kapoor (2007) Controllable Delay-Insensitive Processes, Fundamenta Informaticae 78(1):101-130
Mark Josephs joined the University of Warwick in 2010. He received his BSc in Mathematics from University College, London, and his MSc and DPhil in Computation from the University of Oxford. He is an Honorary Professor of London South Bank University, a Chartered Fellow of the BCS, and a Senior Member of the IEEE. His memberships also include the Editorial Board of the Computer Journal, the EPSRC Peer Review College, and the UK Computing Research Committee.
Tony Tan, University of Edinburgh
Algorithms for static analysis in web databases
Thursday, May 19th 2011. 4pm, CS1.04
The web has brought fundamentally new challenges to data management. The key features that distinguish web data from traditional database applications are its structure: usually described by mark-up languages, such as XML.
The simplest abstraction of XML documents is ordered unranked finite trees whose nodes are labeled by letters from a finite alphabet. This abstraction works well for reasoning about structural properties, but real XML documents carry data, which cannot be captured by a finite alphabet. Thus, there has been a consistent interest in data trees, i.e., trees in which nodes carry both a label from a finite alphabet and a data value from an infinite domain. It seems natural to add at least the equality on data values to a logic over data trees. But while for finitely-labeled trees many logical formalisms -- such as monadic second-order logic -- are decidable; adding data-equality makes even first-order logic undecidable. This explains why the search for decidable reasoning formalisms over data trees has been a common theme in XML research.
In this talk I will give a brief survey on results and techniques that have been developed in the research of data trees and data words. Some of the results have surprisingly deep connections with other well known models of computation such as vector addiction systems and automata with Presburger constraints.
Tony Tan is currently a postdoc (under Leonid Libkin) in the School of Informatics, University of Edinburgh. He got his B.Sc and M.Sc. in National University of Singapore in 2003 and 2005, respectively, and PhD in Technion in 2009.
Nasir Memon, Polytechnic Institute of New York University
Photo Forensics - There is more to a picture than meets the eye
Thursday, May 12th 2011. 11am, CS0.07 (provisional)
Given an image or a video clip can you tell which camera it was taken from? Can you tell if it was manipulated? Given a camera or even a picture, can you find from the Internet all other pictures taken from the same camera? Forensics professionals all over the world are increasingly encountering such questions. Given the ease by which digital images can be created, altered, and manipulated with no obvious traces, digital image forensics has emerged as a research field with important implications for ensuring digital image credibility. This talk will provide an overview of recent developments in the field, focusing on three problems. First, collecting image evidence and reconstructing them from fragments, with or without missing pieces. This involves sophisticated file carving technology. Second, attributing the image to a source, be it a camera, a scanner, or a graphically generated picture. The process entails associating the image with a class of sources with common characteristics (device model) or matching the image to an individual source device, for example a specific camera. Third, attesting to the integrity of image data. This involves image forgery detection to determine whether an image has undergone modification or processing after being initially captured.
Nasir Memon is a Professor in the computer science department at the Polytechnic Institute of New York University, New York. He is the director of the Information Systems and Internet Security (ISIS) lab at Polytechnic (http://isis.poly.edu). Prof. Memon received his BE in Chemical Engineering and MS in Math from BITS, Pilani, India, 1981. He received his MS in Computer Science (1989) and PhD in Computer Science (1992) from the University of Nebraska, Lincoln.
Prof. Memon's research interests include Digital Forensics, Data Compression, Computer and Network Security and Multimedia Computing and Security. He has published more than 250 articles in journals and conference proceedings and holds 6 patents in image compression and security with six more pending application. He has won several awards including the NSF CAREER award and the Jacobs Excellence in Education award. His research has been featured in NBC nightly news, NY Times, MIT Review, Wired.Com, New Science Magazine etc.
He is currently the Editor-in-Chief of the IEEE Transactions on Information Security and Forensics. He was an associate editor for IEEE Transactions on Image Processing, the Journal of Electronic Imaging, the ACM Multimedia Systems Journal, the LNCS Transaction on Data Hiding, IEEE Security and Privacy Magazine, IEEE Signal Processing Magazine and the International Journal on Network Security. Prof. Memon is the co-founder of Digital Assembly (http://www.digital-assembly.com) and Vivic Networks (http://www.vivic.com), two early stage start-ups in NYU-Poly's incubator. He is a fellow of the IEEE and an IEEE Signal Processing Society distinguished lecturer for the years 2011 and 2012.
Shum Prakash (Warwick Ventures) and Olivia Johansson (Venner Shipley LLP)
IP protection and the commercialisation of algorithms and software
Thursday, May 12th 2011. 4pm, CS1.04
Dr Shum Prakash, Business Development Manager, Warwick Ventures, Technology Transfer Office (Tel: 024 7657 4145; Email: firstname.lastname@example.org). After her BSc in Chemistry (Wales), Shum read for her PhD in Biochemistry (Kent) and undertook R&D with Nihon Medi-Physics Co. Ltd and Johnson Matthey Plc to develop radiopharmaceutical drugs for cancer therapy and brain diagnosis. After her MSc in Science and Technology Policy (Sussex), Shum was a consultant at SQW Ltd providing feasibility studies, evaluations and recommendations to many organisations supporting science and technology. She joined Warwick Ventures in December 2001 as a Business Development Manager where she has taken a broad range of technologies from concept to royalty generating commercial licenses and to trading spin off companies. Whilst at Warwick, Shum completed her MBA (Warwick) for which she was awarded a Sainsbury Management Fellowship. She is also a panel member for the RISC fund managed by the NIHR, Department of Health.
Olivia Johansson is a Chartered and European patent Attorney, specialising in electronics, physics and computer implemented inventions at Venner Shipley LLP. Olivia has particular experience in the field of telecommunications and consumer electronics, as well as in satellite technology. Before becoming a patent attorney she worked for a number of university spin-out companies where she developed software. She has particular interest in the fast changing fields of computer memory, processing power and software. Her clients include both multi-national corporations and small start-up companies.
Paul Goldberg, University of Liverpool
The Complexity of the Homotopy Method, Equilibrium Selection, and Lemke-Howson Solutions
Thursday, May 5th 2011. 4pm, CS1.04
Homotopy methods envisage the following general approach to finding a Nash equilibrium of a game. Consider a version of the game where the payoffs have been modified so that there is some obvious Nash equilibrium. Then, continuously change those numbers back towards the values in the game of interest, keeping track of how the Nash equilibrium changes (it should change continuously). The Lemke-Howson algorithm is a very well-known algorithm that can be expressed as a homotopy method.
In the talk, we will see the surprising result that the Lemke-Howson solutions (as well the outputs of other homotopy methods) are PSPACE-complete to compute. This is somewhat paradoxical since the Lemke-Howson method performs very well in practice. This result ties in with earlier work relating the complexity of games, to the complexity class PPAD. But, I will keep the talk reasonably self-contained.
Paul Goldberg is a professor at the department of Computer Science at Liverpool, and head of the Economics and Computation research group. He was previously at the University of Warwick (1997-2006), and before that, at Sandia National Labs (USA) and Aston University. His research currently focuses on equilibrium computation, and models of decentralised interaction of agents.
Martyn Amos, Manchester Metropolitan University
An early warning method for crush
Thursday, March 17th 2011. 4pm, CS1.04
Fatal crush conditions occur in crowds with tragic frequency. Event organizers and architects are often criticised for failing to consider the causes and implications of crush, but the reality is that the prediction and mitigation of such conditions offers a significant technical challenge. Full treatment of physical force within crowd simulations is precise but computationally expensive; the more common method of human interpretation of results is computationally "cheap" but subjective and time-consuming. In this talk we describe an alternative method for the analysis of crowd behaviour, which uses information theory to measure crowd disorder. We show how this technique may be easily incorporated into an existing simulation framework, and validate it against an historical event. Our results show that this method offers an effective and efficient route towards automatic detection of crush.
(Joint work with Peter Harding).
Martyn Amos is a Reader in the School of Computing, Mathematics and Digital Technology at Manchester Metropolitan University, and the Head of the Novel Computation Group. He received his B.Sc. in Computer Science from Coventry University (1993) and his Ph.D. in DNA computing from the University of Warwick (1997). He then held a Leverhulme Special Research Fellowship (University of Liverpool), before permanent academic positions at the Universities of Liverpool (1999-2002) and Exeter (2002-2006). He is currently the Principal Investigator of three projects; the European Commission FP7-funded BACTOCOM (Bacterial Computing) and COBRA (Coordination of Biological and Chemical-IT Research Activities) projects, and the EPSRC-supported Bridging the Gaps: NanoInfoBio project. He has an active interest in public engagement with science, and is the author of the popular science book "Genesis Machines: The New Science of Biocomputing".
Rob Cross, Centre for Mechanochemical Cell Biology, Warwick Medical School
Engines of self organization: mechanochemical cell biology of kinesins and microtubules
Thursday, March 10th 2011. 4pm, CS1.04
Increasing evidence suggests that molecular motors not only pull, push and move along their microtubule tracks, but also modify the assembly dynamics of the micortubule polymers. This creates a potential for patterns of cytoskeletal trafficking to feed back on the structure of the network. My colleagues and I are interested in this and are studying as a model system the regulated dynamics of microtubules in a model cell, the fission yeast S. pombe, using high resolution TIRF and dark field microscopy. Mal3 is an EB-family protein that bintracks the tips of polymerising microtubules and drives towards the A-lattice arrangement of protofilaments, in contrast to the more usual B-lattice. Alp14 and Dis1 are related TOG-family proteins in S. pombe that we find act catalytically to accelerate microtubule growth by up to 10-fold. Klp5 and Klp6 are kinesin-8 family motors whose properties allow them to detect the arrival of microtubule tips at cell ends, and to specifically promote catastrophic depolymerisation of microtubules arriving at cell ends. Our current computational activities are largely in image processing, but we are increasingly interested in developing realistic simulations of complex, dynamic motor-track systems.
Rob Cross is a biophysical cell biologist who works on the mechanisms of bio-molecular motors and the polymerisation dynamics of their tracks. He is professor and director of the newly-established centre for mechanochemical cell biology at Warwick medical school. Rob and his colleagues use techniques in structural biology, transient kinetics, advanced optical microscopy, protein engineering and molecular cell biology to try to understand the machinery of motor-driven spatiotemporal self-organization in biology.
Ingemar Cox, University College London
Probably Approximately Correct Search
Thursday, March 3rd 2011. 4pm, CS1.04
We consider the problem of searching a document collection using a set of independent computers. That is, the computers do not cooperate with one another either (i) to acquire their local index of documents or (ii) during the retrieval of a document. During the acquisition phase, each computer is assumed to randomly sample a subset of the entire collection. During retrieval, the query is issued to a random subset of computers, each of which returns its results to the query-issuer, who consolidates the results. We examine how the number of computers, and the fraction of the collection that each computer indexes, affects performance in comparison to a traditional deterministic configuration. We provide analytic formulae that, given the number of computers and the fraction of the collection each computer indexes, provide the probability of an approximately correct search, where a "correct search'' is defined to be the result of a deterministic search on the entire collection. We show that the randomized distributed search algorithm can have acceptable performance under a range of parameters settings. Simulation results confirm our analysis.
Ingemar J. Cox is currently Professor and Director of Research in the Department of Computer Science at University College London. He is Head of the Future Media Group at UCL.
He has been a recipient of a Royal Society Wolfson Fellowship (2002-2007). He received his B.Sc. from University College London and Ph.D. from Oxford University. He was a member of the Technical Staff at AT&T Bell Labs at Murray Hill from 1984 until 1989 where his research interests were focused on mobile robots. In 1989 he joined NEC Research Institute in Princeton, NJ as a senior research scientist in the computer science division. At NEC, his research shifted to problems in computer vision and he was responsible for creating the computer vision group at NECI. He has worked on problems to do with stereo and motion correspondence and multimedia issues of image database retrieval and watermarking. In 1999, he was awarded the IEEE Signal Processing Society Best Paper Award (Image and Multidimensional Signal Processing Area) for a paper he co-authored on watermarking. From 1997-1999, he served as Chief Technical Officer of Signafy, Inc, a subsidiary of NEC responsible for the commercialization of watermarking. Between 1996 and 1999, he led the design of NEC's watermarking proposal for DVD video disks and later colloborated with IBM in developing the technology behind the joint "Galaxy" proposal supported by Hitachi, IBM, NEC, Pioneer and Sony. In 1999, he returned to NEC Research Institute as a Research Fellow.
He is a Fellow of the IEEE, the IET (formerly IEE), and the British Computer Society. He is a member of the UK Computing Research Committee. He was founding co-editor in chief of the IEE Proc. on Information Security and is an associate editor of the IEEE Trans. on Information Forensics and Security. He is co-author of a book entitled "Digital Watermarking" and its second edition "Digital Watermarking and Steganography", and the co-editor of two books, "Autonomous Robots Vehicles" and "Partitioning Data Sets: With Applications to Psychology, Computer Vision and Target Tracking".
Edmund T Rolls, Department of Computer Science, University of Warwick
How the brain computes and decides
Thursday, February 24th 2011. 4pm, CS1.04
The recognition of objects and faces in natural scenes invariantly with respect to position, size, rotation, and view is a major computational problem not solved in computer vision. How the brain computes will be illustrated by describing a hierarchical neuronal network model that is based on investigations of the functional architecture of the visual system in the brain and that performs these functions. Short-term memory attractor neuronal networks are important for attention in this system, for memory that occurs after object identification, and for decision-making. The operation of these processes enables contrasts to be made between the operation of digital computers and the computations performed by the brain. Understanding the computations performed by the brain is fundamental to understanding the operation of the brain in health and disease. Publications are available at www.oxcns.org.
Professor Edmund T. Rolls, M.A., D.Phil, D.Sc., Hon. D.Sc. is a computational neuroscientist with research interests including the operation of real neuronal networks in the brain in vision, memory, attention, and decision-making; functional neuroimaging of vision, taste, olfaction, feeding, the control of appetite, memory, emotion, and decision-making; neurological disorders of emotion; psychiatric disorders including schizophrenia; and the brain processes underlying consciousness. These studies include investigations in patients, and are performed with the aim of contributing to understanding the human brain in health and disease, and of treating its disorders. He has served as Professor of Experimental Psychology at Oxford University; Fellow and Tutor at Corpus Christi College, Oxford; Vice President, Corpus Christi College, Oxford; Secretary of the European Brain and Behaviour Society; and Secretary to the Council, European Neuroscience Association; and is an Honorary Fellow in the Department of Computer Science. He has published more than 495 full length research papers, which are shown, with many .pdfs available, at http://www.oxcns.org. His books include:
- Rolls,E.T. and Deco,G. (2002) Computational Neuroscience of Vision. Oxford University Press: Oxford.
- Rolls,E.T. (2008) Memory, Attention, and Decision-Making: A Unifying Computational Neuroscience Approach. Oxford University Press: Oxford.
- Rolls,E.T. and Deco, G. (2010) The Noisy Brain: Stochastic Dynamics as a Principle of Brain Processing. Oxford University Press: Oxford.
David Saad, Engineering and Applied Science, Aston University
The statistical physics of noisy computation
Thursday, February 3rd 2011. 4pm, CS1.04
Contributors: Alexander Mozeika, David Saad and Jack Raymond
We show how models of random formulae can be mapped onto a physical framework and employ methods of statistical physics, developed specifically to analyse the typical behaviour of random disordered systems, to gain insight into the behaviour of noisy Boolean random formulae. The stability of the circuit towards input-layer perturbations and its dependence on the input magnetization have been studied to establish the main characteristics of the generated formulae. To investigate the properties of noisy circuits we consider two copies of the same topology with different temperatures representing the noisy and noiseless versions of the same circuit. We show that the typical-case macroscopic behaviour observed corresponds straightforwardly to the bounds obtained in the information theory literature for specific cases. Being very general, the framework is extended to consider further properties of random Boolean formulae for different gates, the level of error and function-bias expected at any depth, the sensitivity to input perturbations and expected convergence rate depending on the input bias, gate properties and gate-noise level. This framework enables one to discover typical properties of noisy computation that are inaccessible via traditional methods of information theory and complements the analysis carried out in the theoretical computer science and information theory communities.
A. Mozeika, D. Saad and J. Raymond, "Computing with Noise - Phase Transitions in Boolean Formulas'', Phys. Rev. Lett. 103, 248701 (2009).
A. Mozeika, D. Saad and J. Raymond, "Noisy Random Boolean Formulae - a Statistical Physics Perspective'', Phys. Rev. E, 82, 041112 (2010).
Professor David Saad holds degrees in both Physics (BA, MSc) & Electronic Engineering (BSc, PhD) from the Technion and Tel Aviv University. He joined the Physics Department at Edinburgh University in 1992 first as a Research Associate and later as a Lecturer. He moved to Aston University in 1995 and was appointed to a Chair in Information Mathematics in 1999. He has over 150 publications, mainly in the application of statistical physics methods and Bayesian statistics to a range of fields; these include neural networks, error-correcting codes, multi-node communication, hard computational problems, network optimisation, advanced inference methods, noisy computation and random Boolean networks. He has held over 15 research grants and is currently heading the Mathematics group at Aston.
Wolfgang Nejdl, Leibniz University Hannover
Web of People -- Improving Search on the Web
Thursday, January 27th 2011. 4pm, CS1.04
More and more information is available on the Web, and the current search engines do a great job to make it accessible. Yet, optimizing for a large number of users, they usually provide good answers only to "most of us", and have yet to provide satisfying mechanisms to search for audiovisual content.
In this talk I will present ongoing work at L3S addressing these challenges. I will start by giving a brief overview of Web Science areas covered at L3S, and the main challenges we adress in these areas, with the Web of People as one important focal point of our research, as well as Web Information Management and Web Search.
In the second part of the talk, I will discuss search for audiovisual content, and how to make this content more accessible. As many of our algorithms focus on exploiting user generated information, I will discuss what kinds of tags are used for different resources and how they can help for search. Collaborative tagging has become an increasingly popular means for sharing and organizing Web resources, leading to a huge amount of user generated metadata. These tags represent different aspects of the resources they describe and it is not obvious whether and how these tags or subsets of them can be used for search. I will present an in-depth study of tagging behavior for different kinds of resources - Web pages, music, and images. I will also discuss how to enrich existing tags through machine learning methods, to provide indexing more appropriate to user search behavior.
Prof. Dr. Wolfgang Nejdl (born 1960) has been full professor of computer science at the University of Hannover since 1995. He received his M.Sc. (1984) and Ph.D. degree (1988) at the Technical University of Vienna, was assistant professor in Vienna from 1988 to 1992, and associate professor at the RWTH Aachen from 1992 to 1995. He worked as visiting researcher / professor at Xerox PARC, Stanford University, University of Illinois at Urbana-Champaign, EPFL Lausanne, and at PUC Rio.
Prof. Nejdl heads the L3S Research Center as well as the Distributed Systems Institute / Knowledge Based Systems , and does research in the areas of search and information retrieval, information systems, semantic web technologies, peer-to-peer infrastructures, databases, technology-enhanced learning and artificial intelligence. Some recent projects in the L3S context include the PHAROS Integrated Project on audio-visual search, the OKKAM IP focusing on entities on the Web, the Digital Library EU project LiWA, coordinated by L3S, which investigates Web archive management and advanced search in such an archive, and the FET IP project LivingKnowledge, which is developing algorithms and methods to handle and exploit diversity, bias and opinion on the Web. Another new project, GLOCAL, focuses on event-based indexing of multimedia data on the web.
Wolfgang Nejdl published more than 250 scientific articles, as listed at DBLP, and has been program chair, program committee and editorial board member of numerous international conferences and journals, most recently including the role of PC chair for WWW'09 in Madrid, PC chair for WSDM'11 in HongKong, and general chair for ICDE'11 in Hannover, see also http://www.kbs.uni-hannover.de/~nejdl/
Simon Miles, King's College London
Automatically Determining the Provenance of Data
Thursday, January 13th 2011. 4pm, CS1.04
The provenance of something, i.e. its history or how it came to be as it is, is important knowledge in many disciplines. In science, for example, the value of a result is in part due to the rigour of the experiment which produced it, while in art collection, the authenticity of an object is judged via records of past ownership. The provenance of computed data (how the data was produced) is also important to know, particularly as computation becomes a more significant part of scientific analyses, medical diagnoses, business decisions etc. The increasing use of large scale distributed systems to process information across organisations, such as in grid computing systems, makes the problem of determining provenance both more difficult and more pressing: it is harder and more important to understand how data was processed while out of your authority. The same is true for information on the web, as data is increasingly copied or cited from one site to another. In this talk, I will give an overview of the problems of determining provenance of data in distributed systems, and the technologies which exist and are being developed to solve them.
Simon Miles is a lecturer in Computer Science at King's College London. He has worked on a range of projects in the areas of e-science, agent-oriented software engineering, electronic contracting, and distributed systems at King's and, previously, at the University of Southampton. He has experience of applying novel open systems technologies to a wide variety of real world use cases, and, as part of his work researching mechanisms for determining the provenance of data, he has collaborated with many international partners. He co-led the first two International Provenance Challenges, bringing together researchers from around twenty disparate teams in two six-month exercises to compare their systems using a single, medical application, and is on the W3C's incubator group for provenance on the web. He has published widely in the areas of multi-agent systems and e-science.
26 November 2010
Cumulative Subgoal Fulfillment: A New Approach to Developing Software
Eric Braude, Boston University
This talk will show how a few principles of physical construction, so self-evident that they are usually unremarked, can be very useful for software construction. This talk will show how the result, formulated as Cumulative Subgoal Fulfillment, applies to classic computer science examples, to Naur's well-known but error-prone text formatting problem, to a video game, to mashups, and to linear programming.
Dr. Braude (Associate Professor of Computer Science; MS, University of Natal; MS, University of Illinois; MS, University of Miami; PhD, Columbia University) teaches software design, artificial intelligence, data structures, information system security, software engineering, and web services. His books have been translated into several languages. Dr. Braude has taught at the University of Pennsylvania, City University of New York, Pennsylvania State University and Seton Hall University, and has served as technology advisor to corporations such as Philips, Lockheed, Lucent Technologies, and MITRE Corporation.
18 November 2010
Raymond Turner, University of Essex
The specification and implementation of computational artifacts occurs within all areas of computer science. Indeed, some would see it as the defining activity of the discipline. Consequently, unpacking its nature, and in particular the nature of specification, should form one of the central themes in the philosophy of computer science. In this talk we address some of the fundamental conceptual questions surrounding the computer science notion of specification.
It is often said that the role of specification is to describe the artifact not how to build it. While this characterization has the merit of being succinct, it hides and suggests a good number of clarifying questions.
1. What is the nature of specification? In particular, do specifications act like definitions? Indeed, do they function in the same way as definitions in mathematics?
2. Are specifications fundamentally different to computer programs? Are programs imperative and specifications descriptive? If so, how do logic, functional and object oriented programs fit into this characterization?
3. What is the relationship between a specification and an artifact? What does it mean to say that an artifact has been correctly constructed? How is correctness expressed and established?
4. What are computational artifacts? What kind of things are they?
5. How do specifications differ from scientific theories?
6. What is the difference between the specification of an artifact and the functional description of it found in a manual for its use?
While some of these issues are implicitly addressed in the relevant computational literature e.g. (Morgan, 1990; Jon 86 : Jones; Van Vliet, 2008), there is little conceptual analysis. Our aim is to contribute to the latter.
4 November 2010
Information theory of communication over multiple correlated noisy channels
Oleg Zaboronski, University of Warwick
Motivated by the example of probe storage we consider the problem of parallel information transmission over multiple communication lines affected by strongly correlated noise. Surprisingly, channel capacity and Gallager's random coding exponent for this channel can be computed analytically. The answers show that a 'traditional' application of error correction coding to this channel will lead to an error floor for ANY code and decoding algorithm.
21 October 2010
Meet the new data model, same as the old data model
Leonid Libkin, University of Edinburgh
In the (very) old days, the world of databases was a big mess, dominated by the network (graph) and the hierarchical (tree) data models. Then Codd came, and the nice and clean relational model replaced all others. In addition to providing a steady employment to many logicians, it created a $20billion/year business. We didn't live in that paradise for too long though: less than 30 years later, the world came back to the hierarchical model (XML). And graph-structured data hasn't been dormant all those years, although it was much less visible than the relational and XML models. Alberto Mendelzon was the first to revisit the graph model back in the 80s, and we have seen more activity lately, due to applications in areas such as RDF, biological databases, and social networks.
In this talk I shall give a few examples showing the importance of graph querying, demonstrate desirable features of graph query languages, and show which problems are easy for graph querying, and which problems suddenly become very hard.
7 October 2010
Modelling non-rigid 3D shapes
Andrew Fitzgibbon, Microsoft Research
I will talk about modelling of 3D objects that can change shape, either in video or from sets of photos. I describe two approaches we have used recently: implicit and explicit 3D, and show how they are related. The implicit approach models the image-formation process as a 2D-to-2D transformation directly from an object's texture map to the image, modulated by an object-space occlusion mask, we can recover a representation which we term the "unwrap mosaic". This representation allows a considerable amount of 3D manipulation without ever being explicitly 3D. The second strand is to explicitly recover 3D, for example from a set of photos. Such a set might be obtained by an image search for the term "clownfish". This yields many photos of clownfish, but no two are of exactly the same individual, nor are they the same 3D shape. Yet, to the human observer, this set of images contains enough information to infer the underlying 3D deformable object class. We aim to recover models of such deformable object classes directly from images. For classes where feature-point correspondences can be found, this is a straightforward extension of nonrigid factorization, yielding a set of 3D basis shapes to explain the 2D data. However, when each image is of a different object instance, surface texture is generally unique to each individual, and does not give rise to usable image point correspondences. We overcome this sparsity using curve correspondences (crease-edge silhouettes or class-specific internal texture edges).
Andrew Fitzgibbon is a principal researcher at Microsoft Research, Cambridge, UK. His research interests are in the intersection of computer vision and computer graphics, with excursions into neuroscience. Recent papers have been on models of nonrigid motion, general-purpose camera calibration, human 3D perception, large-scale image search, and the application of natural image statistics to problems of figure/ground separation and new-view synthesis.
He has co-authored "best papers'' at ICCV (twice), CVPR, and BMVC, and software he wrote won an Engineering Emmy Award in 2002 for significant contributions to the creation of complex visual effects. He studied at University College Cork, at Heirot-Watt university, and received his PhD from Edinburgh University in 1997. Until June 2005 he held a Royal Society University Research Fellowship at Oxford University's Department of Engineering Science.
1 July 2010
The Affordances of Web 2.0/3.0 for Personal Learning Environments
Professor Hugh C. Davis, University of Southampton
Traditionally learning has been seen as a solitary and individualistic task; learning has been represented as committing knowledge to memory and the personal acquisition of skills and literacies. The affordances of early computer technologies amplified this perspective, and transitions of learning technologies to networked platforms sustained the individualist context within the Virtual Learning Environment (VLE). However constructivist critiques of learning environments have emphasised the importance of social interactions and the benefits of groups working and problem solving as a means to learning and knowledge acquisition. Advances in Web technologies over the last decade (the so called Web 2.0) have enabled us to build tools to support and integrate many kinds of collaboration and learning in networks. Such tools have been retrofitted to existing VLEs. This presentation argues that the current generation of Virtual Leaning Environments are no longer fit for purpose; they embody an approach to learning that supports ineffective/inappropriate didactic approaches, and do not complement the expectations or approaches to learning taken by Generation Y learners.
At the University of Southampton, as part of a major curriculum innovation project we have been examining the approaches we wish to take to providing the virtual environments for learning, teaching and research that will be fit for the next ten years. The focus of this presentation will be on our reflections on personal (personalised and personalisable) rich learning environments, and the part played in these environments by Web 2.0, linked data and cloud computing.
Hugh Davis is Professor of Learning Technology within the School of Electronics and Computer Science, where he leads the Learning Societies Lab (http://www.lsl.ecs.soton.ac.uk), a research team of 50-60 people with a focus on Web Technologies and Technology Enhanced Learning. Hugh has a long history of research in the Hypertext domain going back to before the advent of the Web, and has over 200 publications (http://www.ecs.soton.ac.uk/people/hcd/publications) in Hypertext and Learning Technology, as well as numerous grants in these areas. In addition to his research, Hugh is the University Director of Education responsible for technology enhanced learning (TEL) strategy, and this combination of activities enables him to take a research informed and leading edge perspective on the technology that can be delivered at a cross university level.
24 May 2010
Personalisation in the World Wide Web
Professor Vincent Wade, Trinity College Dublin
Personalisation web research have been successful in researching personalisation in niche application areas where web content has been specifically designed for e.g. in Tourist Information Systems, eLearning and Museum systems. More recent years has seen ‘personalisation-lite’ techniques being used by web search providers e.g. personalisation based on country of origin, or on comparison with similar or aggregate queries. However, personal use of the web extends far beyond just personalised content, and encompasses many dimensions of a web experience e.g. personalisation of tasks & activities, personalisation based on cultural preferences and values, and personalisation for social interaction and community engagement etc.
Next generation personalisation is tackling the issues of dynamic adaptation and composition of web content and services drawn from the heterogeneous open web. Such personalisation empowers the user via adaptive experiences combining traditional multimedia web content, end user generated content (wikis, ,blogs, forums, tweets, RSS feeds etc) as well as dynamically customising delivery to devices and multimodal interfaces etc.
In this seminar we will consider new directions and dimensions in personalised, adaptive web and how they can be addressed. We will investigate key challenges involving integrated open corpus & service personalisation, cultural adaptivity (including multilingual personalisation), and indicate how personalisation can enhance the web communities and the wisdom of the crowd. We will explore techniques and technology to enable multi-dimensional web personalisation i.e. personalisation of the web based on multiple, concurrent influences (from individual identity & personal properties, to context, device and service affordances).
Prof Vincent Wade: Vincent is an Associate Professor in the School of Computer Science and Statistics, Trinity College Dublin . He is Director of the Intelligent Systems Discipline which is a cluster of research groups including the Knowledge and Data Engineering Group, the Graphic Vision and Visualisation Group, the Computational Linguistics Group and the Artificial Intelligence Group as well as the Centre for Health Informatics. Vincent is also Deputy Director of the newly formed Centre for Next Generational Localisation and Personalisation (CNGL) sponsored by Science Foundation Ireland. This multi university research centre is focused on the research and development of innovative digital content management, personalization and localization on the web.