Open positions at K.U. Leuven

The machine learning group of the department of computer science of K.U.Leuven has several open positions for PhD students and post-docs, amongst others in the context of the projects
– ERC Starting Grant “MiGraNT: Mining Graphs and Networks, a Theory-based approach”, https://dtai.cs.kuleuven.be/research/projects/ERC2009MiGraNT
– KULeuven OT project “Probabilistic structured models: learning from large-scale hybrid domains”, http://dtai.cs.kuleuven.be/research/projects/OTPSM11
– FWO project “learning from data originating from evolution”, http://dtai.cs.kuleuven.be/research/projects/LDOE09

We are especially looking for candidates interested in one of the following topics:
– Learning the structure of graphical models and probabilistic logical models
– Efficient inference for probabilistic logical models
– Learning in large graphs
– Learning from data originating from evolution
– Learning from processes in relational domains
– Probabilistic logical models for computational biology
– Transfer learning
– Decision support in intensive care
– Chemoinformatics and proteomics

A more detailed description of the topics can be found on http://dtai.cs.kuleuven.be/ml/jobs

The ideal candidate will possess a master degree in computer science, AI, or a closely related discipline, a strong background in computer science and mathematics, a scientific attitude and the ability to reason through problems, excellent programming skills, the ability to communicate written and orally in English in a clear and precise manner, a pro-active and independent attitude as well as the ability to function well in a team environment and a strong motivation.

Interested candidates should apply before August 24th to receive full
consideration, by following the instructions at https://dtai.cs.kuleuven.be/ml/170-jobs-information-for-prospective-ml-phd-students

ICPRAM 2012 Conference – 2nd Call for Papers, deadline extension

2012 International Conference on Pattern Recognition
Applications and Methods (ICPRAM2012)

February 6-8, 2012
Vilamoura, Algarve, Portugal
http://www.icpram.org

ICPRAM (1st International Conference on Pattern Recognition Applications and Methods – http://www.icpram.org/) has an open call for papers, whose deadline is extended to 8 September 2011. We hope you can participate in this conference by submitting a paper reflecting your current research in any of the following tracks:

– Theory and Methods
– Applications

ICPRAM 2012 will be held in Vilamoura, Algarve, Portugal next year, on February 6-8, 2012.

The conference will be sponsored by the Institute for Systems and Technologies of Information, Control and Communication (INSTICC) in cooperation with the Association for the Advancement of Artificial Intelligence (AAAI) and Pattern Analysis, Statistical Modelling and Computational Learning (PASCAL2), and technically co-sponsored by IEEE Signal Processing Society, Machine Learning for Signal Processing
(MLSP) Technical Committee of IEEE, AERFAI (Asociación Espanola de Reconocimiento de Formas y Analisis de Imagenes) and APRP (Associacao Portuguesa de Reconhecimento de Padroes). INSTICC is member of the Workflow Management Coalition (WfMC).

ICPRAM would like to become a major point of contact between researchers, engineers and practitioners on the areas of Pattern Recognition, both from theoretical and application perspectives.
Contributions describing applications of Pattern Recognition techniques to real-world problems, interdisciplinary research, experimental and/or theoretical studies yielding new insights that advance Pattern Recognition methods are especially encouraged.

The conference program features a number of Keynote Lectures to be delivered by distinguished world-class researchers, including those listed below.

The proceedings of ICPRAM will be submitted for indexation by Thomson Reuters Conference Proceedings Citation Index (ISI), INSPEC, DBLP and Elsevier Index (EI).
All accepted papers (full, short and posters) will be published in the conference proceedings, under an ISBN reference, on paper and on CD-ROM support.
A short list of presented papers will be selected so that revised and extended versions of these papers will be published by Springer-Verlag in a AISC Series book.
Top selected papers in specific areas of interest will be published as a special issue in the Neurocomputing Journal.

Best paper awards will be distributed during the conference closing session. Please check the website for further information (http://www.icpram.org/best_paper_awards.asp).

All papers presented at the conference venue will be available at the SciTePress Digital Library (http://www.scitepress.org/DigitalLibrary/).
SciTePress is member of CrossRef (http://www.crossref.org/).

We also would like to highlight the possibility to submit to the following Special Sessions:

– Algebraic Geometry in Machine Learning

– Shape Analysis and Deformable Modeling

– Machine Learning for Sequences

– Pattern Recognition Applications in Remotely Sensed Hyperspectral Image Analysis

– High-Dimensional Inference from Limited Data: Sparsity, Parsimony and Adaptivity

– Interactive and Adaptive Techniques for Machine Learning, Recognition and Perception

Workshops and further special sessions are also invited, with the condition that the topics do not overlap with the existing ones. If you wish to propose a workshop or a special session, for example based on the results of a specific research project, please contact the secretariat. Workshop chairs and Special Session chairs will benefit from logistics support and other types of support, including secretariat and financial support, to facilitate the development of a valid idea.

ICPRAM 2012 will be held in conjunction with ICAART 2012
(http://www.icaart.org/home.asp) and ICORES (http://www.icores.org) in Vilamoura, Algarve, Portugal next year, on February 6-8, 2012.
Registration to ICPRAM will enable free access to the ICAART and ICORES conferences (as a non-speaker).

IMPORTANT DATES:
Regular Paper Submission: September 8, 2011 (extended)
Authors Notification (regular papers): October 17, 2011
Final Regular Paper Submission and Registration: October 31, 2011

CONFERENCE CHAIR:
Ana Fred, Technical University of Lisbon / IT, Portugal

PROGRAM CO-CHAIRS:
J. Salvador Sanchez, Jaume I University, Spain Pedro Latorre Carmona, Jaume I University, Spain

KEYNOTES SPEAKERS:
Ludmila Kuncheva, Bangor University, U.K.
Tiberio Caetano, NICTA, Australia
Francis Bach, INRIA, France
Jose C. Principe, University of Florida, U.S.A.
Joachim M. Buhmann, ETH Zurich, Switzerland

PROGRAM COMMITTEE:
Please check the program committee members at http://www.icpram.org/program_committee.asp

CONFERENCE TRACKS:
TRACK 1: THEORY AND METHODS
– Exact and Approximate Inference
– Density Estimation
– Bayesian Models
– Gaussian Processes
– Model Selection
– Graphical and Graph-based Models
– Missing Data
– Ensemble Methods
– Neural Networks
– Kernel Methods
– Large Margin Methods
– Classification
– Regression
– Sparsity
– Feature Selection and Extraction
– Spectral Methods
– Embedding and Manifold Learning
– Similarity and Distance Learning
– Matrix Factorization
– Clustering
– ICA, PCA, CCA and other Linear Models
– Fuzzy Logic
– Active Learning
– Cost-sensitive Learning
– Incremental Learning
– On-line Learning
– Structured Learning
– Multi-agent Learning
– Multi-instance Learning
– Reinforcement Learning
– Instance-based Learning
– Knowledge Acquisition and Representation
– Meta Learning
– Multi-strategy Learning
– Case-based Reasoning
– Inductive Learning
– Computational Learning Theory
– Cooperative Learning
– Evolutionary Computation
– Information Retrieval and Learning
– Hybrid Learning Algorithms
– Planning and Learning
– Convex Optimization
– Stochastic Methods
– Combinatorial Optimization

TRACK 2: APPLICATIONS
– Natural Language Processing
– Information Retrieval
– Ranking
– Web Applications
– Economics, Business and Forecasting Applications
– Bioinformatics and Systems Biology
– Audio and Speech Processing
– Signal Processing
– Image Understanding
– Sensors and Early Vision
– Motion and Tracking
– Image-based Modelling
– Shape Representation
– Object Recognition
– Video Analysis
– Medical Imaging
– Learning and Adaptive Control
– Perception
– Learning in Process Automation
– Learning of Action Patterns
– Virtual Environments
– Robotics

Call for participation: Challenge on Robust activity recognition @ SMC

Call for participation

ACTIVITY RECOGNITION CHALLENGE
http://www.opportunity-project.eu/challenge

Human activity recognition can be used to devise assistants that
provide proactive support by exploiting the knowledge of the user’s
context, determined from sensors located on-body. Notwithstanding the
large amount of research endeavours on this field, the comparison of
different approaches is often not possible due to the lack of common
benchmarking tools and datasets that allow for replicable and fair
testing procedures across several research groups.

We intend to address this issue by setting up a
challenge on activity recognition addressing key questions in
activity recognition such as classification based on multimodal
recordings, activity spotting and robustness to noise. To this end we
provide a benchmark database of daily activities recorded in a sensor
rich environment.

Prizes will be awarded to participants that achieve the best
performance, and the overall lessons and results obtained from this
challenge will be presented at the associated workshop at the IEEE
Conference on Systems, Man and Cybernetics 2011. Moreover, we are
currently arranging the future publication of selected contributions in
a top journal in the field.

More information: http://www.opportunity-project.eu/challenge
Contact: activityrecognition.challenge(at)gmail.com

Important dates
—————
September 9, 2011: Final submission date
October 9, 2011: Final results and conclusions presented at the SMC workshop

Organisers
———-
Ricardo Chavarriaga, EPFL, Switzerland
Daniel Roggen, ETHZ, Switzerland
Alois Ferscha, Johannes Kepler University, Linz, Austria
Paul Lukowicz, U. Passau, Germany

Two PhD positions in Computer Vision at the University of Edinburgh

Applications are invited for two fully funded PhD students to work in
the School of Informatics on the following topics:

* knowledge transfer to automate learning visual models
* learning visual object categories from consumer and advertisement videos
* leveraging the structure of natural sentences to aid visual learning

Applicants must have:

* Master degree (preferably in Computer Science or Mathematics)
* Excellent programming skills; the projects involve programming
in Matlab and C++
* Solid knowledge of Mathematics (especially algebra and statistics)
* Highly motivated
* Fluent in English, both written and spoken
* UK or EU nationality
* Experience in computer vision and/or machine learning is a plus
(ideally a master thesis in a related field)

The School of Informatics at Edinburgh is one of the top-ranked
departments of Computer Science in Europe and offers an exciting
research environment. Edinburgh is a beautiful historic city with a
high quality of life.

Starting date: January 2012 or later

The PhD work will be carried out under the supervision of Vittorio
Ferrari. He is currently with ETH Zurich. He will move to the
University of Edinburgh in December 2011 and build a new research
group in Computer Vision. For an overview of his current research
activities, please visit

http://www.vision.ee.ethz.ch/~calvin/

For pre-screening, please send applications to the email address
below, including:
* complete CV
* title and abstract of your master thesis
* complete grades for all exams passed during both the bachelor and master
(to obtain this position you need high grades, especially in
mathematics and programming disciplines)
* the name and email address of one reference (preferably your
master thesis supervisor)
* if you already have research experience, please include a publication list.

email: vittorio.ferrari(at)ed.ac.uk

Three Post-Doctoral Research Assistantships and PhD Studentships

The UK-EPSRC funded project “Learning to Recognise Dynamic Visual Content from Broadcast Footage” is a collaboration between the University of Surrey (Prof Richard Bowden), the University of Oxford (Prof Andrew Zisserman) and the University of Leeds (Dr Mark Everingham) with research staff appointed at each institution.

The aim of the proposed research is to develop computer vision and machine learning methods for understanding dynamic visual content. These methods will support systems capable of automatically recognising from video the actions of people, and their interactions with others, objects and the environment. There are two demonstration activities in the areas of Sign Language Recognition and that of more general broadcast content. The project will develop new approaches to weakly supervised learning from subtitles and transcripts.

The project is funded for four years and each institution will host one RA and there are also funded PhD positions available, however all parties will work closely together to fulfil the project objectives.

Potential candidates for the postdoctoral position should hold a PhD in a related discipline with a proven track record in terms of international publications. You should have a research background in computer vision or machine learning. Experience in areas such as human detection, tracking and Sign Language, gesture or activity recognition would also be an advantage. You should have the technical skills necessary to create new and exploit existing software tools for video analysis.

Informal academic enquiries may be made to: Prof Richard Bowden, +44 (0)1483 689838,

r.bowden(at)surrey.ac.uk.

PASCAL Visual Object Classes Recognition Challenge 2011 (VOC2011) – Training & test data available

We are running the PASCAL Visual Object Classes Recognition Challenge again this year. As in 2010 there are 20 object classes for the main competitions. Participants can recognize any or all of the classes, and there are classification, detection and pixel-wise segmentation competitions. This year the action classification taster competition has a new “other” category, and there is also a taster competition on person layout (detecting head, hands, feet). There is also an associated large scale visual recognition taster competition organized by www.image-net.org.

We remind users that the PASCAL VOC Challenge evaluation server is available at http://host.robots.ox.ac.uk:8080/ and is currently providing evaluations and downloads for the VOC 2008, 2009, 2010 and 2011 challenges.

The development kit (Matlab code for evaluation, and baseline algorithms) and training/test data is now available at:

http://pascallin.ecs.soton.ac.uk/challenges/VOC/voc2011/index.html

where further details are given. The timetable of the challenge is:

* 25 May 2011: Development kit (training and validation data plus
evaluation software) made available.

* 14 July 2011: Test set made available.

* 10 October 2011. Deadline for submission of results.

* 7 November 2011: Workshop in association with ICCV 2011, Barcelona.

Mark Everingham
Luc Van Gool
Chris Williams
John Winn
Andrew Zisserman

Open PhD position in Marseille, France

Spectral methods for learning tree-structured graphical models.

Supervision : François Denis and Liva Ralaivola, Université
d’Aix-Marseille

Deadline : 08/20/2011

A PhD studentship is available as part of the french ANR funded project
LAMPADA on “Learning Algorithms, Models an sPArse representations for
structured DAta” being jointly undertaken by Inria Lille Nord Europe (Marc
Tommasi), the Laboratoire d’Informatique Fondamentale de Marseille
(François Denis), the Laboratoire Hubert Curien de Saint-Etienne (Marc
Sebban) and the Laboratoire d’Informatique de Paris 6 (Patrick Gallinari).
See http://lampada.gforge.inria.fr/wiki….

The student will be based in the Laboratoire d’Informatique Fondamentale
de Marseille (France) and join the QARMA team headed by Liva Ralaivola.
He/she will be supervised by Pr François Denis
(http://www.lif.univ-mrs.fr/ fdenis) and Liva Ralaivola
(http://www.lif.univ-mrs.fr/ liva). The project will include collaboration
with the project MOSTRARE (Lille INRIA Nord Europe).

The studentship is funded by an ANR project and will start from 1st
October 2011 or as soon as possible thereafter. The studentship is funded
for 3 years (currently 1400 euros per month – net income).

Requirements : The PhD candidate should have or expect to obtain a MSc or
equivalent in computer science or mathematics. The following qualities are
desirable : strong interests in machine learning or statistics ; excellent
record of academic and/or professional achievement ; strong mathematical
skills ; strong programming skills ; good written and spoken communication
skills in French or English.

Project Description :

Graphical models are probabilistic models where dependencies between
observable and latent random variables are represented by a graph. HMM are
simple kinds of graphical models. Tree-structured graphical models extend
the expressivity of HMM and are widely used in bio-informatics and Natural
Language Processing. The question of the inference of parameters of a
tree-structured graphical model is not completely solved, even when the
tree topology of the underlying model is unique and supposed to be known.
The inference of the topology of the model from samples is still an open
question.

Recent learning methods rely on algebraic and geometrical properties of
these models and reduce the inference of parameters topology of the models
to spectral computations on matrices. Indeed, tree-structured graphical
models can be seen as particular rational tree series which can be
described by means of natural algebraic operators.

At first, the student will study the link between two formalisms,
tree-structured graphical models and tree rational stochastic languages,
and compare the algorithms described in the references below. Then, she/he
will study the question of the identifiability of tree-structured
graphical models from samples and explore the possibility of developing a
semi-supervised-like learning algorithm, when some topology-samples pairs
are available. Two extensions could be considered : kernelization and
extension to more general graphical models, for example based on regular
graphs.

References :

– A Spectral Algorithm for Learning Hidden Markov Models, Daniel Hsu, Sham
M. Kakade, Tong Zhang, COLT 2009
– A Spectral Algorithm for Latent Tree Graphical Models, Parikh, Le Song,
Xing, ICML 2011
– Grammatical inference as a principal component analysis problem, R.
Bailly, F. Denis, and L. Ralaivola, ICML 2009
– A Spectral Approach for Probabilistic Grammatical Inference on Trees, R.
Bailly, A. Habrard, F. Denis, ALT 2010
– A Hilbert Space Embedding for Distributions, Smola, Gretton, Le Song,
Schölkopf, ALT 2007 – Nonparametric Tree Graphical Models via Kernel
Embeddings, Le Song, Gretton, Guestrin, AISTAT 2010
– Learning Latent Tree Graphical Models de Choi, Tan, Anandkumar et
Willsky, JMLR 2011

=== Contact and application ===

For further information, please contact Prof. François Denis
(francois.denis(at)lif.univ-mrs.fr), or Prof. Liva Ralaivola
(liva.ralaivola(at)lif.univ-mrs.fr).

If you are interested and believe that you qualify, please send your
application to Prof. François Denis and Prof. Liva Ralaivola, before
August 20, 2011. Include :
– Curriculum Vitae with the names and contact details of at least 2
references
– A list of exams and grades obtained
– A cover letter explaining how your skills and research interests fit the
project

Data Mining School in Maastricht, The Netherlands

9-th Summer School on Dat Mining, Maastricht, The Netherlands
http://www.unimaas.nl/datamining/
An intensive 4-day introduction to methods and applications

Department of Knowledge Engineering, Maastricht University,
Maastricht, The Netherlands
August 29 – September 1, 2011

Introduction
Most business organizations collect terabytes of data about business
processes and resources. Usually these data provide just “facts and
figures”, not knowledge that can be used to understand and eventually
re-engineer business processes and resources. Scientific community in
academia and business have addressed this problem in the last 20 years
by developing a new applied field of study known as data mining.
In practice data mining is a process of extracting implicit,
previously unknown, and potentially useful knowledge from data. It
employs techniques from statistics, artificial intelligence, and
computer science. Data mining has been successfully applied for
acquiring new knowledge in many domains (like Business, Medicine,
Biology, Economics, Military, etc.). As a result most business
organizations need urgently data-mining specialists, and this is
the point where this school comes to help.

Description
The school curricullum is well balanced between theory and practice.
Each lecture is accompanied by a lab in which participants experiment
with the techniques introduced in the lecture. The lab tool is Weka, one
of the most advanced data-mining environments. A number of real data
sets will be analysed and discussed. In the end of the school
participants develop their own ability to apply data-mining techniques
for business and research purposes.

Tools
The school focuses on techniques with a direct practical use.
A step-by-step introduction to powerful (freeware) data-mining tools
will enable you to achieve specific skills, autonomy and hands-on
experience. A number of real data sets will be analysed and discussed.
In the end of the school you will have your own ability to apply data-
mining techniques for research purposes and business purposes.

Content
The school will cover the topics listed below.
– The Knowledge Discovery Process
– Data Preparation
– Basic Techniques for Data Mining:
+ Decision-Tree Induction
+ Rule Induction
+ Instance-Based Learning
+ Bayesian Learning
+ Support Vector Machines
+ Regression Techniques
+ Clustering Techniques
+ Association Rules
– Tools for Data Mining
– How to Interpret and Evaluate Data-Mining Results

Intended Audience
This school is intended for four groups of data-mining beginners:
students, scientists, engineers, and experts in specific fields who need
to apply data-mining techniques to their scientific research, business
management, or other related applications.

Prerequisites
The school does not require any background in databases, statistics,
artificial intelligence, or machine learning. A general background in
science is sufficient as is a high degree of enthusiasm for new
scientific approaches.

Certificate
Upon request a certificate of full participation will be provided after
the school.

Registration
To register for the school please send an email to:

smirnov(at)maastrichtuniversity.nl

In the e-mail please specify:
– Name
– University / Organisation
– Address
– Phone
-E-Mail

Registration Deadline: August 22, 2011

Registration fees
Academic fee 600 Euros
Non-academic fee 850 Euros

Included in the price are: school material and coffee breaks. The local
cafeteria will be available for lunch (not included).

Registration e-mail: smirnov(at)maastrichtuniversity.nl

Regular mail should be sent to:

Evgueni Smirnov
Department of Knowledge Engineering
Faculty of Humanities and Sciences
Maastricht University
P.O.Box 616
6200 MD Maastricht
The Netherlands
Phone: +31 (0) 43 38 82023
Fax: +31 (0) 43 38 84897
E-mail: smirnov(at)maastrichtuniversity.nl

The PASCAL Probabilistic Inference Challenge (PIC11) – Call for Submissions

Probabilistic inference in graphical models is a key task in many applications, from machine vision to computational biology. Since the
problem is generally computationally intractable many approximations have been suggested over the years.

The goal of this challenge is to evaluate inference algorithms on difficult large scale problems.
Some of challenge highlights are:
– Algorithms for MAP, marginals, and partition function approximation will be evaluated.
– Solvers will be evaluated on different time scales (20 seconds, 20 minutes, 1 hour).
– An online leaderboard will show the relative rankings of the algorithms.
– Modest cash prizes will be available on all categories.
– After the challenge the evaluation site will remain active as a service to the community.

Timeline:
July 15th, 2011: Site open to registration and submission of solvers.
October 1st, 2011: Challenge Begins. No new models added after this date.
December 31st, 2011: Site closed to submissions.
January 15th, 2012: Winners announced.

Organizers: Gal Elidan and Amir Globerson (Hebrew University).
Web master: Uri Heinemann (Hebrew University)

Web Page: http://www.cs.huji.ac.il/project/PASCAL/

PhD Studentship – Learning to Recognise Dynamic Visual Content from Broadcast Footage

Supervision: Dr Mark Everingham, University of Leeds

Deadline: Open until filled

Project Description:

A PhD studentship is available as part of an EPSRC funded project on “Learning to Recognise Dynamic Visual Content from Broadcast Footage” being jointly undertaken by the University of Leeds (Mark Everingham), the University of Oxford (Andrew Zisserman), and the University of Surrey (Richard Bowden).

The objective of the project is to develop automated tools that allow temporal visual content, such as a human gesturing, using sign language, or interacting with objects or other humans, to be learnt from standard TV broadcast signals using the transmitted annotation in the form of subtitles and annotation for the visually impaired as supervision. This requires the development of models of the visual appearance and dynamics of actions, and learning methods which can train such models using the weak supervision provided by the subtitles. Once the models have been learnt they can then be used without supervision, e.g. for sign language interpretation or automatic description of the content of video footage, and during the project demonstrators will be engineered for both of these applications.

The student will focus on the development of visual descriptors and learning algorithms for sign language and action recognition in broadcast video. S/he will be based in the School of Computing at the University of Leeds, and will be supervised by Dr Mark Everingham (http://www.comp.leeds.ac.uk/me/).

Funding Notes: The studentship is funded by an EPSRC project studentship and will start from 1st October 2011 or as soon as possible thereafter. The studentship is funded for 3 years and covers Home/EU fees and maintenance at the standard EPSRC rate (currently £13,590 per annum). Applications are welcome from overseas students, but such students would have to provide the difference between the UK/EU and the overseas student rates for university fees from some other source, such as a scholarship or personal funds.

Academic Staff Contact Details: Dr Mark Everingham. M.Everingham(at)leeds.ac.uk (for informal enquiries about the project only – do not send applications to this address).

Entry Requirements: The PhD candidate should have or expect to obtain a first class or strong 2.1 honours degree in computer science, mathematics, or related discipline. The following qualities are desirable: demonstrable experience in computer vision or machine learning; excellent record of academic and/or professional achievement; strong mathematical skills; strong programming skills, especially C/C++ and Matlab; good written and spoken communication skills in English.

Details on how to apply can be found at:

http://www.leeds.ac.uk/rds/prospective_students/apply/I_want_to_apply.html

Application Procedure: Formal applications for research degree study must be made either on line through the University website, or on the University’s application form. Detailed information of how to apply on line can be found at: http://www.leeds.ac.uk/students/apply_research.htm

The paper application form is available at: http://www.leeds.ac.uk/rsa/prospective_students/apply/I_want_to_apply.html

Please return the completed application form to the Research Degrees & Scholarships Office, University of Leeds, LS2 9JT.

Please note, if you intend to send academic references we can only accept them if they are on official letter headed paper and contain an original signature and stamp; they must arrive in sealed envelopes. Alternatively, the School will contact your named academic referees directly.