Announcements








July 10-14, 2017

ISIPTA 17 & ECSQARU 2017

The Tenth International Symposium on Imprecise Probability: Theories and Applications (ISIPTA 17) and Fourteenth European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty (ECSQARU 2017) will be held jointly in Lugano (Switzerland), on July 10-14, 2017.

Both the ISIPTA symposia and the ECSQARU conferences are biennial events to discuss theory and practice of reasoning under uncertainty. While ISIPTA is especially focused on the field of imprecise probabilities, the scope of ECSQARU covers more generic qualitative and numeric uncertainty paradigms. The co-location of the two events is intended to favor cross-fertilization among researches active in both communities.


July 4-9, 2017

Fourth Summer School on Belief Functions and Their Applications (BELIEF 2017)

The 4th Summer School on Belief Functions Theory and Applications was hosted by the School of Automation, Northwestern Polytechnical University (NPU), Xi'an, P.R. China, on July 5-9, 2017.
The 20th International Conference on information Fusion (Fusion 2017) was jointly held in Xi'an, on July 10-13, 2017.

Belief function theory, also known as Dempster-Shafer theory (DST), provides an interesting framework to represent and combine uncertain information. This summer school focussed on the following topics: belief functions and its applications in pattern recognition, information fusion, and the related topics like rough sets, etc. Its purpose is to provide opportunities to exchange ideas for the researchers from institutes and Universities all over the world and present new results on the theory of belief functions and related areas.


Deadline: December 1st, 2016

Postdoctoral Researcher in Statistical Machine Learning

Salary: around 28K. This is a one-year post, further extension subject to funding.

The Department of Computing and Communication Technologies is seeking to appoint a Postdoctoral Research Assistant in Statistical Machine Learning in the Artificial Intelligence Laboratory, in order to kick-start a new research project on novel robust foundations for machine learning.

Machine learning algorithms typically focus on fitting a model to the available training data ('overfitting'), which may lead, for instance, an autonomous driving system to perform well on validation tests but fail catastrophically when tested in the real world (as it has unfortunately been demonstrated of late). Common practice in the field contemplates ‘generalisation’ criteria which are based on a rather naïve correlation between smoothness and generality.
With the deployment of machine learning algorithms in critical AI systems, however, it is crucial to ensure that these algorithms behave predictably 'in the wild'. This raises issues with Vapnik's traditional Statistical Learning Theory as a viable foundation for robust machine learning. The remarkable empirical performance achieved so far by deep learning-based approaches, on the other hand, is yet to be matched by a satisfactory understanding of the theoretical basis for their robust behaviour.

The successful candidate will take the leadership of a new research programme designed at laying the groundwork for a new, robust paradigm for the foundations of machine learning. A number of research avenues can be envisaged. Worst-case, cautious predictions may be generated by solving appropriate mini-max optimisation problems. Robust analyses based on convex sets of models (e.g. convex sets of linear boundaries) appear promising. Finally, robust Bayesian analysis and random set theory may provide the means for a generalisation of the concept of “Probably Approximately Correct” in statistical learning theory.

Candidates should have a PhD or other Postgraduate qualification or be studying for PhD in a relevant subject and have significant experience in machine learning, statistics or both.
The successful candidate will join a vibrant and ambitious department that is welcoming, supportive and friendly. The department blends excellence in teaching and knowledge transfer with world-leading research in areas that span Artificial Intelligence, Computer Vision, Cognitive Robotics, Augmented Reality, Wireless Communications and Human Machine Interfaces.

Closing date: December 1st, 2016


July 12 2015

UAI 2015 Tutorial PDF - Belief functions for the working scientist - VIDEO NOW ONLINE!

The theory of belief functions, sometimes referred to as evidence theory or Dempster-Shafer theory, was first introduced by Arthur P. Dempster in the context of statistical inference, to be later developed by Glenn Shafer as a general framework for modelling epistemic uncertainty. The methodology is now well established as a general framework for reasoning with uncertainty, with well-understood connections to related frameworks such as probability, possibility, random set and imprecise probability theories. Importantly, in recent years the number of papers published on the theory and application of belief functions has been booming (reaching over 800 in 2014 alone), displaying strong growth in particular in the East Asian community and among practitioners working on multi-criteria decision making, earth sciences, and sensor fusion.

Belief functions are a natural tool to cope with heavy uncertainty, lack of evidence and missing data, and extremely rare events. An early debate on the rationale of belief functions gave a strong contribution to the growth and success of the UAI community and series of conference in the Eighties and Nineties, thanks to the contribution of scientists of the caliber of Glenn Shafer, Judea Pearl, Philippe Smets and Prakash Shenoy, among others. Ever since the UAI and BELIEF community have somewhat diverged, and the proposers� effort has been recently directed towards going back to a closer relationships and exchange of ideas between the two communities.

This was one of the aims of the recent BELIEF 2014 International Conference of which the proposers were General Chair and member of the Steering Committee, respectively. A number of books are being published on the subject as we speak, and the impact of the belief function approach to uncertainty is growing. The proposed tutorial aims at bridging the gap between researchers in the field and the wider AI and Uncertainty Theory community, with the longer term goal of a more fruitful collaboration and dissemination of ideas.


Deadline: May 1 2015

Computer Vision Development Engineer (KTP Associate)

Salary: Up to �30k pa depending on qualifications and experience, plus generous training budget. This is a fixed-term post for 24 months.

This challenging post for a computer vision specialist needs creative thinking to generate original inspection solutions that will shape the future technology of a world-leading company.

Meta Vision Systems, located close to Oxford, UK, has designed a project in collaboration with Oxford Brookes University and the partners are seeking a highly motivated, Master�s qualified individual to lead this development over two years.

Based at the company but working under supervision from the university, you will have access to people and facilities in both organisations, including the Oxford Brookes AI and Vision group. You will also work with the company�s clients, designing customised algorithms to meet their requirements.

Meta Vision Systems works internationally with major companies in sectors as diverse as automotive, aerospace, robotics and general fabrication, supplying its innovative laser vision systems for material joining. On successful completion of the project the company expects to offer a permanent position.

You will have carried out Master�s level research in computer vision or machine learning, and you will be proficient in coding. You will benefit from a generous training budget and as part of the long-established Knowledge Transfer Partnerships programme you will have access to hundreds of others engaged in KTP projects across the UK.

For information see www.meta-mvs.com, http://cct.brookes.ac.uk and http://ktp.innovateuk.org.
To apply, see www.brookes.ac.uk/vacancies Ref: 436/19288/BC

Closing date: 1 May 2015 Interview date: 14 May 2015


Deadline: February 27 2015

Special Session on "Belief Functions" at ECSQARU 2015

David Mercier and Fabio Cuzzolin are happy to announce the upcoming Special Session on belief functions at the ECSQARU 2015 International Conference.
The deadline is approaching fast: FEBRUARY 27, 2015

Description and covered topics

Belief function theory, also known as Dempster-Shafer theory or evidence theory, is a well established general framework for reasoning with uncertainty, with well understood connections to other frameworks such as probability, possibility and imprecise probability theories. First introduced by Arthur P. Dempster in the context of statistical inference, the theory was later developed by Glenn Shafer into a general framework for modeling epistemic uncertainty. These early contributions have been the starting points of many important developments, including the Transferable Belief Model and the Theory of Hints. Belief function theory is now getting more and more applied in many research areas.

This special session is then dedicated to belief functions regarding theoretical and/or applications as well as practical aspects when using belief functions. This moment will be an opportunity to exchange new ideas and new results on work based on the belief function framework.
Original papers with theoretical and/or practical contributions are solicited.
Possible topics include, but are not limited to:

Theoretical aspects
  • Independence
  • Conditioning
  • Combination rules
  • Continuous belief functions
  • Graphical models
  • Decision making
  • Inference
  • Statistical estimation

Applications
  • Data fusion
  • Diagnostic
  • Classification
  • Pattern recognition
  • Clustering
  • Tracking
  • Data mining
  • Signal and image processing
  • Semantic web
  • Medical diagnosis
  • Business decision
Organizers
  • Fabio Cuzzolin, Reader, Head of Artificial Intelligence and Vision, Oxford Brookes University, United Kingdom.
  • David Mercier, Associate Professor, University of Artois, France.

Paper Submission

Papers to the special session should be submitted through the general paper submission website as regular submissions. During submission, authors submitting to the special session should check the corresponding category. Papers submitted to the special session will undergo the same review and decision process as regular papers.

The submission website is

https://ecsqaru2015.hds.utc.fr/submission/article/submission?lang=en

Important Dates

Submission and decision dates are the same as regular submissions.


Deadline: October 31 2014

PhD Studentship - "Real-time Action Recognition for Human-Robot Interaction"

The Faculty of Technology, Design and the Environment at Oxford Brookes University is pleased to offer a three year full-time PhD Studentship to a new student commencing in January 2015. The successful applicant will receive an annual bursary of �10,000 for three years (with no inflation increase) and the fees will be paid by the University.

The successful candidate will work within the Department of Computing and Communication Technologies � as part of both the Artificial Intelligence and Vision group and the Cognitive Robotics group, under the supervision of Dr Fabio Cuzzolin and Dr Nigel Crook.

Topic of research: Real-time Action Recognition for Human-Robot Interaction

Action recognition is a fast-growing area of research in computer vision. The problem consists in, given a video captured by one or more cameras, detecting and recognising the category of the action performed by the person(s) who appear in the video. The problem is very challenging, for a number of reasons: labelling videos is an ambiguous task, as the same sequence can be assigned different verbal descriptions by different human observers; different motions can carry the same meaning (inherent variability); nuisance factors such as viewpoint, illumination variations, occlusion (as parts of the moving person can be hidden behind objects or other people) further complicate recognition. In addition, traditional action recognition benchmarks are based on a �batch� philosophy: it is assumed that a single action is present within each video, and videos are processed as a whole, typically via algorithms which require entire days to be completed. This can be ok for tasks such as video browsing and retrieval over the internet (although speed is a huge issue there), but is completely unacceptable for a number of real world applications which require a prompt, real-time interpretation of what is going on. Examples are: human-robot and human-machine interaction (using gestures to send commands to a computer or a robot), surveillance (detecting potentially dangerous actions or events in live feeds), car driver�s monitoring (monitoring the level of attention, or responding to gestural commands), gaming (interpreting the body language of a video game player), intelligent vehicles (understanding the behaviour of pedestrians and other vehicles in the vicinity of a car). Consequently, a new paradigm of �online�, �real-time� action recognition is rapidly emerging, and is likely to shape the field in coming years. The AI and Vision group is already building on its multi-year experience in batch action recognition to expand towards online recognition, based on two distinct approaches: one based on the application of novel �deep learning� neural networks to automatically segmented video regions, the other resting on continually updating an approximation of the space of �feature� measurements extracted from images, via a set of balls of radius which depends on how difficult classification is within that region of the space.

For further information about the Artificial Intelligence and Vision group, consult http://cct.brookes.ac.uk/research/isec/artificial-intelligence/ and Dr Fabio Cuzzolin�s web page: http://cms.brookes.ac.uk/staff/FabioCuzzolin/
For information on the Cognitive Robotics group, consult http://cct.brookes.ac.uk/research/isec/cognitive-robotics/index.html

The selection criteria will focus on academic excellence, suitability of research experience and skills, subject knowledge and references.
If you would like to apply you should request an application pack from Ms Zane Kalnina, tdestudentships@brookes.ac.uk, quoting �Flood risk PhD studentship� in the subject line.

Fully completed applications must be sent to tdestudentships@brookes.ac.uk - by Friday 31th October 2014.

As part of the application you must submit a maximum 3 page outline research proposal (instructions on how to prepare a research proposal can be found on the Brookes website: http://cct.brookes.ac.uk/research/proposals.html , together with a supporting statement of no more than 300 words summarising: � Your reasons for undertaking this project; � Preparation undertaken and previous research experience; � How this bursary will make a difference to your research.

Please be advised that the selection process may involve an interview for a selected list of candidates, and the successful candidate would be expected to commence in the research degree programme in January 2015.


Deadline: February 28 2014

PhD Studentship - "Uncertainty in Computer Vision"

The Department of Computing and Communications Technologies at Oxford Brookes University is pleased to offer a three year full-time PhD Studentship to a new student commencing in June 2014. The successful applicant will receive an annual bursary of �10,000 for three years (with no inflation increase) and the fees will be paid by the Universty.
The successful candidate will work within the Artificial Intelligence and Vision group of the Department of Computing and Communication Technologies, under the supervision of Dr Fabio Cuzzolin.

Topic of research: Uncertainty in Computer Vision

Decision making and estimation are central in most applied sciences, as the need often arises to make inferences about the state of the external world, based on information which is at best limited, if not downright misleading. Uncertainty can be dealt with in a number of ways. Generative probabilistic graphical models, which describe how the data are generated via classical distribution functions, are most used for complex, multi-person activity recognition. Discriminative models which do not attempt to model data generation, but focus on learning how to discriminate between data belonging to different categories or classes, are dominant in action and gesture recognition, in which we aim at recognising human actions based on a limited training set of examples, captured via conventional or range cameras. Imprecise-probabilistic models which assume the data is probabilistic but insufficient to estimate a precise probability distribution, have been successfully employed in example-based human pose estimation. Depending on the problem we need to tackle, we might need to consider metric learning techniques for generative models, latent SVM part-based discriminative approaches, or a meaningful integration of the two.

The successful candidate will work on both the theoretical development and the application of these techniques to scenarios such as the interaction with a humanoid robot able to recognise and mimic natural human gesturing, the retrieval of videos from internet repositories such as YouTube, the monitoring of the health of people affected by brain conditions in their own homes, via range sensors such as Kinect.

For further information about the Artificial Intelligence research group, consult

http://cct.brookes.ac.uk/research/isec/artificial-intelligence/index.html

or Dr Fabio Cuzzolin�s web page:

http://cms.brookes.ac.uk/staff/FabioCuzzolin/

The selection criteria will focus on academic excellence, suitability of the research environment for your project and references.

If you would like to apply you should submit an application for a place on the PhD programme at Oxford Brookes University through UKPASS (http://www.ukpass.ac.uk). As part of the application you must also submit a full research proposal (instructions on how to prepare a research proposal can be found on the Brookes website: http://cct.brookes.ac.uk/research/proposals.html) together with a supporting statement of no more than 500 words summarising:

Your reasons for undertaking this project; Preparation undertaken and previous research experience; The ways in which this bursary will make a difference to your research.

Please apply through UKPass and submit all of the following supporting documents separately to Helen Tanner - htanner@brookes.ac.uk - by 12noon on 28 February 2014.

Full research proposal
Copy of Passport (if applicable)
IELTS Certificate or equivalent (if applicable)
Two academic references
Degree Certificate(s)
Transcript(s)

Please be advised that the selection process may involve an interview, and the successful candidate would be expected to commence in the research degree programme in June 2014.


January 2014

Article on Research Media "Innovation International" on Fabio's project on tensorial models for gait identification

Fabio has published an article on Research Media's "International Innovation" magazine on his EPSRC project "Tensorial modeling of dynamical systems for gait and activity recognition".

International Innovation published by Research Media is the leading global dissemination resource for the wider scientific, technology and research communities, dedicated to disseminating the latest science, research and technological innovations on a global level. More information and a complimentary subscription offer to the publication can be found at: www.researchmedia.eu

Full article in high resolution PDF format


September 26-29, 2014

The 3rd International Conference on Belief Functions - Call for Papers

The theory of belief functions, also referred to as evidence theory or Dempster-Shafer theory, was first introduced by Arthur P. Dempster in the context of statistical inference, and was later developed by Glenn Shafer as a general framework for modeling epistemic uncertainty. These early contributions have been the starting points of many important developments, including the Transferable Belief Model and the Theory of Hints. The theory of belief functions is now well established as a general framework for reasoning with uncertainty, and has well understood connections to other frameworks such as probability, possibility and imprecise probability theories.

In 2012 alone, more than 300 papers on belief functions and their applications have been published worldwide. The ambition of the series of International Conferences on Belief Functions - BELIEF - is to bring together the large and expanding community of mathematicians, statisticians, computer scientists, engineers, economists and practitioners which work on the theoretical foundations of belief calculus or its application to all fields of applied science.

This 3rd edition, in particular, aims at more closely involving in the community the many specialists of other fields which use belief functions in their daily work, improving the overall visibility of the field by pushing towards a more coalesced and tightly connected community, and reaching out towards the sibling fields of uncertainty theory, Bayesian reasoning, imprecise probability and fuzzy theory.

The conference will provide opportunities to exchange ideas and present new results on both the theory and applications of belief functions and related areas such as random sets, imprecise probability and possibility theory.
Original contributions are solicited on theoretical aspects, including:
  • decision making
  • combination rules
  • conditioning
  • continuous belief functions
  • independence and graphical models
  • statistical inference
  • geometry and distance metrics
  • mathematical foundations
  • computational frameworks
as well as on applications in various areas including, but not limited to:
  • data and information fusion
  • pattern recognition
  • machine learning and clustering
  • tracking and data association
  • data mining
  • signal and image processing
  • computer vision
  • medical diagnosis
  • business decision
  • risk analysis
  • engineering and environment
  • climatic change
Papers will be presented orally during the conference in a single track session, or in poster sessions.

Full paper submission deadline: April 30th, 2014 (see IMPORTANT DATES).

Authors of selected papers from the BELIEF 2014 conference will be invited to submit an extended version of their papers for possible inclusion in a special issue of the International Journal of Approximate Reasoning.


Deadline: January 16 2013

PhD Studentship in Belief Theory

Supervisor: Dr Fabio Cuzzolin

The Department of Computing and Communications Technologies at Oxford Brookes University is pleased to announce that it is able to offer a Doctoral Bursary to a new PhD student in for full time study commencing in January 2013. The successful applicant will have course fees waived and will be awarded a bursary of �10,000 per annum for three years (with no inflation increase).

The PhD project will be focus on decision making and estimation. These are central problems in most applied sciences, as both people and machines need to make inferences about the state of the external world, and take appropriate actions. Traditionally, the (uncertain) state of the world is assumed to be described by a probability distribution over a set of alternative, disjoint hypotheses.
Sometimes, however, as in the case of extremely rare events (e.g., a volcanic eruption), few statistics are available to drive the estimation. Part of the data can be missing. Furthermore, under the law of large numbers, probability distributions are the outcome of an infinite process of evidence accumulation, drawn from an infinite series of samples: in all practical cases, instead, the available evidence only provides some sort of constraint on the unknown probabilities governing the process. All these issues have led to the recognition of the need for a coherent mathematical theory of uncertainty.
Shafer�s theory of belief functions, in particular, allows us to express partial belief by providing lower and upper bounds to probability values. It is appealing because it addresses all the above mentioned issues; its rationale is neat and simple; it is a straightforward generalization of probability theory; it does not require abandoning the notion of event. The widespread influence of uncertainty at different levels explains why belief functions are being increasingly applied to fields as diverse as robotics, fault analysis, decision making, sensor fusion, machine vision, and many more.
Mathematically, a belief function is a random set, i.e. a probability distribution on the power set (the collection of all subsets). Equivalent alternative interpretations can be given in terms of compatibility relations, inner measures, and sum functions. However, the necessary mathematical tools for prediction and estimation in this framework have only partially been developed yet, due to their inherent complexity: this is the case, in particular, for the generalization of the classical total probability theorem to belief functions.

The aim of this studentship is to study the mathematical properties of belief functions, and give a contribution towards the development of crucial tools such as the total belief theorem.

Informal requests: fabio.cuzzolin@brookes.ac.uk Formal applications: follow the instructions here:

http://www.jobs.ac.uk/job/AFL096/phd-studentship/

and contact jheaton@brookes.ac.uk


Deadline: August 31 2012

Postdoctoral Research Assistant in gait and activity recognition

Supervisors: Dr Fabio Cuzzolin
Faculty/Directorate: Technology, Design and Environment
Department: Department of Computing and Communication Technologies
Title of post: Postdoctoral Research Assistant in gait and activity recognition
Grade of post: 7
Post number: 18353
FT or % P/T: FT
Permanent/Temporary: Temporary
Principal location of work: Wheatley campus
Immediate line manager: Dr Fabio Cuzzolin
Staff managed: None
Qualifications required for post: Ph.D. in Computer Science or related fields.
Experience required for post: Substantial experience in computer vision or machine learning research, as a doctoral student. An adequate research record in computer vision or machine learning. A potential for conducting research independently. Coding proficiency in C++ and Matlab.
Overall purpose of post: To undertake research activity within the framework of an EPSRC grant on �Tensorial modelling of dynamical systems for gait and activity recognition�.


April 2012

The 3rd International Conference on Belief Functions

Fabio has been chosen to be the Chair of the upcoming BELIEF 2014 International Conference on Belief Functions, Theory and Applications. The conference will take place in Oxford in Spring 2014.


November 1-2 2011

GEOMIP - First Workshop on the Geometry of Imprecise Probabilities

The main aim of this workshop is to discuss and study aspects of the interaction between geometry and imprecise probability and related statistical methods. The meeting room is on the top floor of the Calman Learning Center, opposite to the Department of Mathematical Sciences.
The workshop is held at Durham University, Department of Mathematical Sciences. Contact Frank Coolen for more information.


Deadline: January 27 2011

PhD Studentship in Computer Vision/Robotics - Intelligent Transport Systems

Supervisors: Dr Fabio Cuzzolin and Professor Phil Torr
Eligibility: The following full-time studentship which includes bursary and fees is available for a maximum of 3 years. Applicants require a good Honours degree (2.1 or equivalent) and Home, EU and International students are eligible.
Start date: June �2011
Value p.a.: �13,500 bursary & fees (International Students); �14,500 bursary & fees (EU Students)
Closing Date: 27 January 2011
The School of Technology at Oxford Brookes University is funding a PhD studentship in Computer Vision and Robotics/Autonomous Navigation under the new Intelligent Transport Systems Doctoral Training Programme.

Autonomous navigation of intelligent vehicles requires solving a number of sophisticated vision problems, such as the recognition of static (signs, cars) and moving (people, animals, bicycles) obstacles along the vehicle�s path or in its vicinity, the localization and tracking of 3D points in dynamic scenes, object motion prediction and recognition. Techniques derived from recent developments in manifold learning, graph optimization and spectral decomposition may be potentially useful in tackling such problems.

Informal enquiries: Dr Fabio Cuzzolin (fabio.cuzzolin@brookes.ac.uk), Professor Phil Torr (philiptorr@brookes.ac.uk)

Further information about the research: http://tech.brookes.ac.uk/research/intelligent-transport-systems
Instructions for applicants can be found on our website: http://tech.brookes.ac.uk/research/intelligent-transport-systems/studentships To apply for this studentship, please complete the application forms available at the web address above.


Deadline: February 4 2011

ISIPTA '11: 7th International Symposium on Imprecise Probability: Theories and Applications - Call for Papers

The ISIPTA meetings are the primary international forum to present and discuss new results on the theories and applications of imprecise probability.
Imprecise probability is a generic term for the many mathematical and statistical models and methods, allowing us to measure chance or uncertainty without the restriction of sharp probabilities. These models include lower and upper expectations or previsions, interval valued probabilities, sets of probability measures, belief functions, Choquet capacities, comparative probability orderings, possibility measures, plausibility measures, and sets of desirable gambles. Imprecise probability models are needed in inference and decision problems where the relevant information is scarce, vague or conflicting, and where preferences may be incomplete.

ISIPTA '11 will be held in Hotel Grauer B�r in Innsbruck, Austria. More information about Hotel Grauer B�r can be found on this website: www.innsbruck-hotels.at/hotel-grauer-baer/das-hotel/.


Deadline: January 31 2011

Eleventh European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty - Call for Papers

The biennial ECSQARU conferences are a major forum for advances in the theory and practice of reasoning under uncertainty. Contributors are expected to come both from researchers interested in advancing the technology and from practitioners who are using uncertainty techniques in applications. The scope of ECSQARU includes, but is not limited to, fundamental issues, representation, inference, learning, and decision making in qualitative and numeric paradigms.

The proceedings of ECSQARU 2011 will be published in the Springer Lecture Notes in Artificial Intelligence series. Authors are requested to prepare their conference papers in the LNCS/LNAI format. Papers must not exceed 12 pages and must be submitted as PDF electronically.

This year, the best papers from the conference will be selected and presented at a special track of IJCAI 2011 conference, and revised versions be included in the IJCAI 2011 proceedings. The conference invited speakers are: Professor D Dubois, Professor D Gabbay, and Professor J Halpern.