Ph.D. Position in Machine Learning at INRIA Lille – Team SequeL

Applications are invited for a Ph.D. studentship on the general area of “Sequential Decision-making under Uncertainty” at INRIA Lille – Team SequeL. Below is the detail of this call.

Title: Sequential Decision-Making with Big Data

Keywords: sequential decision-making, reinforcement learning, learning and planning in MDPs and POMDPs, exploration/exploitation dilemma, bandit algorithms, adaptive resource allocation, regret minimization, optimization

Research Program:

The candidate is expected to conduct research on both theoretical and applied aspects of the problem of “Sequential Decision-making with Big Data” (see the description below), collaborate with researchers and Ph.D. students at INRIA and outside, and publish the results of her/his research in conferences and journals. The candidate will work with Mohammad Ghavamzadeh (http://chercheurs.lille.inria.fr/~ghavamza) and other researchers at Team SequeL (https://sequel.lille.inria.fr).

This Ph.D. program is focused on the problem of dealing with big data and limited resources in sequential decision-making under uncertainty.

– Big Data: Sequential decision-making applications that need to handle Big Data can be classified into three categories, which define related research problems.

1) Very large number of data points: This is a typical case in time series data that are fairly simple, but sampled at high frequency, such as user clicks on the web and financial data. In this scenario, the most important issue is the computational cost.
2) Very high-dimensional input space: This case arises when each data point consists of a lot of measurements, leading to a curse of dimensionality. Examples are customer information in online marketing problems and problems with complex sensors (such as Kinect cameras). The best way to solve this type of problem is to leverage intrinsic regularities (e.g., smoothness, sparsity, dependencies in features) to reduce the dimensionality.
3) Partially observable input space: Often, the observed input measurements do not have sufficient information for accurate decision-making, but one can leverage the history of the observations to improve the situation. This often requires projecting the problem into a high-dimensional representation.

– Limited Resources: In many real-world sequential decision-making applications we only have a limited budget of resources such as number of samples or access to a system’s simulator etc. When the available resources (sample or computation) are limited and/or access to more resources is costly, it would be absolutely necessary to allocate the available resources (or ask for more resources) efficiently in order to find good strategies. The problem of adaptive resource allocation has been studied in bandits, planning, and stochastic optimization, but there still exist many open problems and challenges in this area that require further investigation.

– Other Related Problems that arise in real-world applications of sequential decision-making: (i) how to evaluate a policy learned from a batch of historical data (generated with a different policy) with minimum interaction with the real-world environment, (ii) learning risk-sensitive and robust strategies, (iii) learning interpretable policies (i.e., policies that are understandable by experts of the problem at hand, who do not necessarily know much about machine learning, like medical doctors or financial managers) etc.

Requirements:

The applicant will have a Master’s (or equivalent) degree in Computer Science, Statistics, or related fields, with background in reinforcement learning, bandit algorithms, statistics, and optimization. Programming skills will be considered as a plus. The working language of the group is English, so the candidate is expected to have good communication skills in English.

About INRIA and Team SequeL:

SequeL (https://sequel.lille.inria.fr) is one of the most dynamic teams at INRIA (http://www.inria.fr), with over 25 researchers and Ph.D. students working on several aspects of machine learning from theory to application, including statistical learning, reinforcement learning, and sequential decision-making. The SequeL team is involved in national and European research projects and has collaboration with international research groups. This allows the Ph.D. candidate to collaborate with leading researchers in the field at top universities in Europe and North America such as University College of London (UCL), University of Alberta, and McGill University. Lille is the capital of the north of France, a metropolis with over one million inhabitants, and with excellent train connection to Brussels (30min), Paris (1h) and London (1h30).

Benefits:

– Duration: 36 months – starting date of the contract : October 2013, 15th
– Salary: 1957.54 Euros the first two years and 2058.84 Euros the third year
– Monthly salary after taxes: around 1597.11 Euros the first two years and 1679,76 Euros the 3rd year (benefits included)
– Possibility of French courses
– Help for housing
– Participation for transportation
– Scientific Resident card and help for husband/wife visa

Application Submission:

The application should include a brief description of the applicant’s research interests and past experience, plus a CV that contains her/his degrees, GPAs, relevant publications, name and contact information of up to three references, and other relevant documents. Please send your application to mohammad.ghavamzadeh@inria.fr. The deadline for the application is April 15 but the applicants are encouraged to submit their application as soon as possible.

This call has also been posted on

1) my webpage at

http://chercheurs.lille.inria.fr/~ghavamza/phd-ad-2013.html

2) the INRIA website at:

http://www.inria.fr/institut/recrutement-metiers/offres/theses/campagne-2013/%28view%29/details.html?id=PGTFK026203F3VBQB6G68LONZ&LOV5=4509&LG=FR&Resultsperpage=20&nPostingID=7222&nPostingTargetID=12647&option=52&sort=DESC&nDepartmentID=10

Last call for abstracts – ROKS 2013 – Leuven July 8-10, 2013

ROKS-2013

International workshop on advances in Regularization, Optimization, Kernel methods and Support vector machines: theory and applications

July 8-10, 2013, Leuven, Belgium
http://www.esat.kuleuven.be/sista/ROKS2013

SCOPE

One area of high impact both in theory and applications is kernel methods and support vector machines. Optimization problems, learning and representations of models are key ingredients in these methods. On the other hand considerable progress has also been made on regularization of parametric models, including methods for compressed sensing and sparsity, where convex optimization plays a prominent role. The aim of
ROKS-2013 is to provide a multi-disciplinary forum where researchers of different communities can meet, to find new synergies along these areas, both at the level of theory and applications.

The scope includes but is not limited to:
– Regularization: L2, L1, Lp, lasso, group lasso, elastic net, spectral regularization, nuclear norm, others
– Support vector machines, least squares support vector machines, kernel methods, gaussian processes and graphical models
– Lagrange duality, Fenchel duality, estimation in Hilbert spaces, reproducing kernel Hilbert spaces, Banach spaces, operator splitting
– Optimization formulations, optimization algorithms
– Supervised, unsupervised, semi-supervised learning, inductive and transductive learning
– Multi-task learning, multiple kernel learning, choice of kernel functions, manifold learning
– Prior knowledge incorporation
– Approximation theory, learning theory, statistics
– Matrix and tensor completion, learning with tensors
– Feature selection, structure detection, regularization paths, model selection
– Sparsity and interpretability
– On-line learning and optimization
– Applications in machine learning, computational intelligence, pattern analysis, system identification, signal processing, networks, datamining, others
– Software

INVITED SPEAKERS

Francis Bach, INRIA
Stephen Boyd, Stanford University
Martin Jaggi, Ecole Polytechnique Paris
James Kwok, Hong Kong University of Science and Technology Yurii Nesterov, Catholic University of Louvain UCL Massimiliano Pontil, University College London Justin Romberg, Georgia Tech John Shawe-Taylor, University College London Alexander Smola, Google & UC Berkeley Joel Tropp, California Institute of Technology Ding-Xuan Zhou, City University of Hong Kong

CALL FOR ABSTRACTS

The ROKS-2013 program will feature invited plenary talks, oral sessions and poster sessions. Interested participants are cordially invited to submit an extended abstract (max. 2 pages) for their contribution. After the workshop a number of selected contributions will be invited for an edited book.

For further information see http://www.esat.kuleuven.be/sista/ROKS2013 .

IMPORTANT DATES

– Deadline extended abstract submission: March 4, 2013
– Notification of acceptance: April 8, 2013
– Deadline for registration: June 3, 2013
– International Workshop ROKS-2013: July 8-10, 2013

ORGANIZING COMMITTEE

Chair: Johan Suykens (KU Leuven)

Andreas Argyriou (Ecole Centrale Paris), Kris De Brabanter (KU Leuven), Moritz Diehl (KU Leuven), Kristiaan Pelckmans (Uppsala University), Marco Signoretto (KU Leuven), Vanya Van Belle (KU Leuven), Joos Vandewalle (KU Leuven)

Co-sponsored by ERC Advanced Grant

ICC’2013 – International Create Challenge

http://www.createchallenge.org
Call for participation open
!! Create your startup in 3 weeks !!

September 21 – October 11, 2013
Martigny, Switzerland

Description and Objectives
—————————
The goal of the 2013 International Create Challenge (ICC’2013) is to foster the creation of start-ups within the framework of Human & Media Computing. The ICC’2013 is an initiative supported by the National Centre of Competence in Research (NCCR) on Interactive Multimodal Information Management (IM2, www.im2.ch), via its association (AIM2), and the Idiap Research Institute (Idiap, www.idiap.ch).

The ICC’2013 is a free of charge 3-week immersive technology transfer accelerator program giving entrepreneurs the unique opportunity to develop their original idea towards a “Minimum Viable Product’’ (e.g., demonstrator, product prototype) in collaboration with groups of entrepreneurs and researchers.

The ICC’2013 combines the availability of state-of-the-art technologies, cutting edge research, mentor-led coaching, and micro-seed investment. The global value of the endowments and awards (cash as well as in-kind rewards) amounts to more than 200’000 CHF. The ICC’2013 can accommodate up to 30 participants, ideally split into 10 teams of 3 people.

Application
————-
By opening this call for participation, AIM2 and Idiap seek to attract highly motivated “entrepreneurs’’ to create or join a team, develop their original idea towards a “Minimum Viable Product’’, eventually resulting in the creation of a company.

Participants can apply as the representative of a group (to be briefly described) or as individuals interested in joining an initiative.

The call is split into two successive competitive steps. More information is available at http://www.createchallenge.org:

• Call for participation: http://www.createchallenge.org/icc-2013/call-for-participation-2013.pdf/view
• Terms and Conditions: http://www.createchallenge.org/icc-2013/terms-and-conditions-icc-2013.pdf/view
• Apply now: http://www.createchallenge.org/icc-2013/apply-now

• LinkedIn group: http://www.linkedin.com/groups/ICC-International-Create-Challenge-4386640/about

Important dates
——————
• Step 1 Application deadline: 14 June 2013 (11:59pm CET)
• Step 1 Notification of decision: 21 June 2013
• Step 2 Submission deadline: 26 July 2013 (11:59pm CET)
• Step 2 Notification of decision: 16 August 2013
• ICC’2013 Start: 21 September 2013
Closing ceremony: 11 October 2013

Contact and information
————————-
Questions should be directed to icc@idiap.ch.

Research Scientist in Statistical Natural Language Processing at Xerox

The Parsing & Semantics research area at Xerox Research Centre Europe
(XRCE) is currently looking for an experienced researcher in statistical natural language processing (NLP), with a deep understanding of machine learning and/or information extraction (e.g. event extraction). The ideal candidate would also have experience or knowledge of textual entailment, knowledge representation, and combining machine learning with expert knowledge. The applicant should have good coding skills (e.g. Java programming), with the ability to develop research prototypes and pilots.

The successful candidate will be expected to identify challenging problems, develop new solutions, and work with business and development teams to ensure that these solutions have a significant impact. We work together with top academic partners and expect our researchers to publish results in top-tier conferences and journals. We also have multiple open innovation collaborations with academic partners world-wide.

The Parsing & Semantics group concentrates on automatically making sense of electronic documents using semantic analysis. The group focuses on natural language processing methods for robust parsing, semantic analysis, and information discovery, including the role of context in determining meaning. We are particularly interested in theoretical models of communication, language, computation, learning and inference which take into account the context in which these activities occur. The Parsing & Semantics group collaborates closely with the Machine Learning for Services group and the Machine Learning for Document Access and Translation group. We are also interested in applying research results to practical applications and real-world problems. Our general application focus is on converting unstructured text into structured information. The solutions we develop are expected to play a key role in Xerox’ next generation document and business process outsourcing services in domains such as customer care, healthcare, and financial services.

XRCE is located in Grenoble, France, in the heart of the French Alps.
Grenoble offers an excellent quality of life and a large scientific community. For more information, please see http://www.xrce.xerox.com/.

Requirements:

* PhD in Computer Science or Computational Linguistics
* NLP knowledge and experience
* Knowledge or experience in machine learning or information extraction
* Object oriented programming skills (e.g. java)
* Strong written and oral communications skills in English

Application instructions:

The application deadline is March 1, 2013, but applications will be
considered beyond this date until the position is filled.

Informal inquiries can be made to James.Henderson@xrce.xerox.com or
Tonya.Love@xerox.com.
To submit an application, please send your CV and cover letter to both
xrce-candidates@xrce.xerox.com and to Tonya.Love@xerox.com. You should
also include in your CV at least three referees we can contact for
letters of recommendation.

TPAMI Special Issue : Deadline April 1st 2013

CALL FOR PAPERS – TPAMI Special Issue on “Higher Order Graphical Models in Computer Vision: Modelling, Inference & Learning”

This special issue invites paper submissions addressing the issues of modelling, inference, and learning in models with higher-order terms and global constraints. We also welcome survey and overview papers in these general areas.

The call for papers is available at:
(pdf) http://www.computer.org/cms/Computer.org/transactions/cfps/cfp_tpamisi_hogmcvmil.pdf
(html) http://www.di.ens.fr/~alahari/tpami-cfp.html.

The timeline is as follows:

Submission Deadline: April 1, 2013
Reviews: October 1, 2013
Revisions of Submissions: January 1, 2014
Final Decisions/Manuscript: April 1, 2014
Estimated Online Publication: Fall 2014

Regards,
Karteek Alahari

[on behalf of the guest editors:
Karteek Alahari (INRIA-WILLOW / ENS)
Dhruv Batra (TTI-Chicago / Virginia Tech)
Srikumar Ramalingam (MERL)
Nikos Paragios (Ecole Centrale de Paris / Ecole des Ponts-ParisTech / INRIA-Saclay)
Rich Zemel (University of Toronto)]

KTP Associate – Applied Machine Learning – Newcastle University

Salary: up to £27,900 per annum
Closing date: 22 March 2013
Reference: D1181R

You will work on the application of cutting edge machine learning and computational intelligence methods to real world industrial problems. In particular, we are interested in the combined use of support vector machines and unsupervised clustering methods to analyse high dimensional heterogeneous industrial data. The aim of the analysis is to improve the precision of the manufacturing process of printed circuit boards. The only company in the world that offers a fully automated solution to this problem is the XACT PCB Ltd based near Newcastle in the North East of England. The size of the market of printed circuit boards is in the range of billions of dollars. You can have a real impact on how printed circuit boards are made by joining this project.
The project is a collaboration between the XACT PCB Ltd and Newcastle University. You will work most of the time at the offices of the company and every week you will spend at least a half day at the School of Computing Science of Newcastle University where you will also have an office place. The School of Computing Science recently expanded its academic staff with interests in applied machine learning and computational intelligence. Ongoing research includes the analysis of human behavioural data recorded by movement sensors, development of intelligent and adaptive living environments, data mining of bioinformatics databases, analysis of neuroinformatics imaging data, optimisation of gene regulation for synthetic biology, and experimental validation of cyber-physical systems. You will work under the supervision of Dr Peter Andras (peter.andras@ncl.ac.uk).

We are looking for a self-motivated person with a PhD on a topic related to machine learning or computational intelligence (the actual PhD area can be computer science, mathematics, physics, statistics, engineering or any other related field). You should have exceptionally strong skills in developing and coding machine learning algorithms (preferably in C# or other similar languages, including Java, C++, Matlab, R). You should have a clear desire to move towards industry and to make a real world impact through top quality research.

You must have a First Class honours degree or a Distinction level MSc degree in Computer Science, Mathematics, Physics or related fields and preferably a PhD in one of these fields. You will have the experience in software development, large-scale data analytics, development and application of machine learning methods, together with a positive attitude and good interpersonal, communication and team working skills.

The position comes with benefits including a £4,000 individual training budget and management training.

The post is fixed term for a duration of 2 Years.

To apply go to www.ncl.ac.uk/vacancies/ and search for the vacancy with reference D1181R.

Please note this position is subject to the confirmation of funding. Applicants are expected to be contacted by April 2013.

XACT is the world’s leading provider of integrated registration solutions to many of the world’s highest technology PCB plants.

For further details about the XACT PCB Ltd please see www.xactpcb.com
Newcastle University is one of the top UK universities, member of the select Russell Group formed by the 24 leading UK universities. The School of Computing Science is one of the top research departments in the UK in the area of computer science with a research budget of over £4 million per year. The School is a partner in the Newcastle Culture Lab which is a leading UK hub for innovative cultural and social applications of digital technology.

For further details about Newcastle University please visit our information page at http://www.ncl.ac.uk/about/

For further details on the School of Computing Science please see http://www.ncl.ac.uk/computing/

For further details about Knowledge Transfer Partnerships please visit our Research and Enterprise Service webpage at: http://www.ncl.ac.uk/res/knowledge/ktp/index.htm

ECML/PKDD 2013: second call for tutorials

The ECML-PKDD 2013 Organizing Committee invites proposals for half-day tutorials to be held on the first and last days of the conference, which will take place in Prague, Czech Republic from September 23th to 27th, 2013.

Tutorials are intended to provide a comprehensive introduction to core techniques and areas of interest for the machine learning and the data mining community. We are interested in tutorials on established or emerging research topics in the areas above but we also welcome tutorials from related research fields or applications. The ideal tutorial should attract a wide audience. It should be broad enough to provide a basic introduction to the chosen research area, but it should also cover the most important topics in depth. Each tutorial should be well-focused so that its content can be covered in a half day slot (3.5h including a 30-minute break). Proposals that exclusively focus on the presenter’s own work or commercial presentations are strongly
discouraged.

Tutorial slides will be made available online at the main ECML-PKDD server, although authors can provide them in additional websites as well.

Tutorial speakers will be offered the possibility of a reduced or waived registration fee, depending on the number of speakers for the tutorial.

Submission details
• A title and abstract of the tutorial.
• A brief description of the tutorial content and its relevance to the ECML PKDD community (no more than 2 pages).
• A brief outline of the tutorial structure showing that the tutorial’s core content can be covered in a 3.5 hours slot (including a coffee break).
• The names, postal addresses, phone numbers, and email addresses of the tutorial instructors, including one-paragraph statement of their research interests and areas of expertise.
• A list of previous venues and approximate audience sizes, if the same or a similar tutorial has been given elsewhere; otherwise an estimate of the audience size.
• A description of special requirements for technical equipment (e.g., internet access).

Proposals should be submitted by electronic mail, in plain ASCII text no later than March 8, 2013 to: tutorials@ecmlpkdd2013.org

The subject line should be: “ECML PKDD 2013: TUTORIAL PROPOSAL“.

The proponents will receive an email confirmation that their proposal has been received. If no confirmation arrives within 24 hours after sending the proposal, the proponents should contact the tutorial co-chairs again.

Proposal review
The proposal will be reviewed by the tutorial co-chairs, who may use the help of external reviewers, expert on the submission topics.
The features that will be evaluated are:
i. the interestingness for the ECML PKDD areas, which should result in a large audience;
ii. the clarity of the tutorial, which should emerge from its description;
iii. good organization as appearing from the outline;
iv. the adequacy of the speakers, i.e., her/his background/experience in teaching the target topics; and
v. the ability to explain the topics to a large audience with heterogeneous background.
Tutorial speaker responsibilities

Accepted tutorial speakers will be notified by March 29, 2013, and must then provide the revised abstracts (ASCII format) of their tutorials by April 5, 2013 for advertisement purposes in the conference website. Tutorial speakers must provide their tutorial materials (pdf format), at least containing copies of the course slides as well as a bibliography for the material covered in the tutorial, by September 1, 2013.

Important dates
• Tutorial proposal deadline: Friday, March 8, 2013
• Tutorial acceptance notification: Friday, March 29, 2013
• Tutorial abstract due: Friday, April 5, 2013
• Tutorial slides due: Sunday, September 1, 2013

Contacts
In case you have any question, please do not hesitate to contact us. We are looking forward to your proposals,
Sofus Macskassy (University Southern California)
Kurt Driessens (University of Maastricht)
(The ECML/PKDD 2013 Tutorial co-chairs)
tutorials@ecmlpkdd2013.org

POSTDOC IN MACHINE LEARNING AND PROBABILISTIC MODELING, DL 28 FEB 2013 3:00 PM EET

The Finnish Center of Excellence in Computational Inference Research (COIN), a joint research center of Aalto University and University of Helsinki, is now seeking postdoctoral researchers to join us in developing and applying new methods of machine learning and probabilistic modeling, together with professors Samuel Kaski and Petri Myllymäki, with opportunities for collaboration with other professors at COIN. There are opportunities for both methods development and theoretical work, and interdisciplinary applications.

Deadline 28 February 2013 at 3:00 p.m. EET. Please see more details of the application procedure at http://research.ics.aalto.fi/coin/vacancies.shtml

The Finnish Center of Excellence in Computational Inference Research
(COIN) works on fundamental questions of inference and in applications in Intelligent Information Access, Computational Molecular Biology and Medicine, Computational History, Computational Climate, Computational Neuroscience and other directions yet to be determined.

More information:
http://research.ics.aalto.fi/coin/index.shtml (COIN) http://www.aalto.fi/en (Aalto University) http://www.helsinki.fi/research (University of Helsinki)

Funded PhD position in Sequential Learning with Similarities, SequeL team, INRIA, Lille, France

Funded PhD position in Sequential Learning with Similarities, SequeL team, INRIA, Lille, France

Description: The goal of this PhD position is to design and analyze efficient algorithms for decision making under uncertainty. Specifically, this position will focus on cases when the possible actions are somehow related. This extra information can be provided in a form of weighted graph or a similarity metric. The aim is to provide provably optimal algorithms that can be applied in large-scale scenarios, such as movie recommendation or social networks. The purpose is to minimize feedback that we need to give the algorithm in order to make it “intelligent”. In other words, we want the decision-making algorithms that converge fast to nearly-optimal solutions with minimal feedback (samples). For this purpose, we often need to learn about the environment (explore) at the same time as choosing the currently most promising option (exploit). This requires careful allocation of resources, which could be financial costs, CPU time, or a human effort.
Keywords: machine learning, graph-based learning, exploration-exploitation tradeoff, bandit algorithms, learning with similarities, minimal feedback
Objectives: The multi-armed bandit problem is a simple model to study the trade-off between exploration and exploitation. While simple enough, it displays most of the issues that arise in a variety of decision-making problems under uncertainty. The following problems are the most pertinent to the proposed research program:
• Contextual bandits: This setting is an extension of the multi-armed bandit problem where the best decision depends on the information (context) that is repeatedly given to the decision maker [S09]. Selecting relevant web advertisement or the news feed recommendations are the most common applications of this setting [LCLS10].
• Bandits on Graphs: In many problems, the similarities between the decisions are provided in the form of a graph that relates the pairs of the decisions (nodes), potentially with a weight. This can refer to connections in the social networks [CKLB12] or more general combinatorial problems [BL12, YM11]. Another use is in the recommender systems where we want to discover user preferences (assumed to be smooth on a given graph) with minimal number of queries.
• Stochastic optimization: The optimization of a noisy objective function is a very general problem whose difficulty strictly depends on the properties of the function itself (e.g., linear [AHR08], Lipschitz [BMSS08], submodular [HK12]) and the space the function is defined on (e.g., finite support, continuous). What other interesting settings are (functions and spaces) and what set of assumptions is really needed to successfully optimize the function are issues currently under investigation.
Job Description: The PhD candidate will focus on one or more issues related to the problem of learning and decision-making under uncertainty with some similarity information. The PhD candidate will first acquire expertise in different topics of machine learning such as online learning, multi-armed bandit, statistical learning theory, and graph-based learning. The candidate is then expected to contribute to the advancement of the literature on this problem along many different lines: methodological (e.g., definition of general abstract models for a wide range of decision-making problems), theoretical (e.g., near optimality performance guarantees), and algorithmic (e.g., development of novel algorithms for specific decision-making problems). The candidate will work with Michal Valko (http://researchers.lille.inria.fr/~valko/), Remi Munos, and other members of the lab. The applicant will also have the opportunity to collaborate with researchers in several countries in Europe and USA.

Requirements: The successful candidate will have a MSc or equivalent degree in computer science with a strong background in theory or in mathematics. Programming skills will be considered a plus. The working language in the lab is English, a good written and oral communication skill are required.
• Application closing date: Spring 2013
• Duration: 3 years (a full time position)
• Starting date: October 1st, 2013
• Supervisors: Michal Valko and Remi Munos
• Place: SequeL, INRIA Lille – Nord Europe
About INRIA: SequeL (https://sequel.lille.inria.fr/) is one of the most dynamic labs at INRIA (http://www.inria.fr/), with over 25 researchers and PhD students working on both fundamental and practical aspects of sequential learning problems: from statistical learning, through reinforcement learning, to games. Lille is the capital of the north of France, a metropolis with 1 million inhabitants, with excellent train connection to Brussels (30 min), Paris (1h) and London (1h30). Established in 1967, Inria is the only public research body fully dedicated to computational sciences. Combining computer sciences with mathematics, Inria’s 3,400 researchers strive to invent the digital technologies of the future. Educated at leading international universities, they creatively integrate basic research with applied research and dedicate themselves to solving real problems, collaborating with the main players in public and private research in France and abroad and transferring the fruits of their work to innovative companies. The researchers at Inria published over 4000 articles a year. They are behind over 270 active patents and 105 start-ups. The 171 project teams are distributed in eight research centers located throughout France.

Benefits:
• Duration: 36 months – starting date of the contract : October 2013, 15th
• Salary: 1957,54 € the first two years and 2058,84 € the third year
• Salary after taxes: around 1597,11€ the 1st two years and 1679,76 € the 3rd year (benefits included).
• Possibility of French courses
• Help for housing
• Participation for public transport
• Scientific Resident card and help for husband/wife visa
References:
• [AHR08] Jacob Abernethy, Elad Hazan, and Alexander Rakhlin. Competing in the Dark: An Efficient Algorithm for Bandit Linear Optimization. In Proceedings of the 21st Annual Conference on Learning Theory (COLT’08), 2008.
• [BL12] N. Cesa-Bianchi and G. Lugosi Combinatorial bandits Journal of Computer and Systems Sciences, 78:1404-1422, 2012.
• [BMSS08] S. Bubeck and R. Munos and G. Stoltz and Cs. Szepesvari. Online Optimization of {X}-armed Bandits. In Proceedings of the Advances in Neural Information Processing Systems (NIPS’08), 2008.
• [BMS09] S. Bubeck and R. Munos and G. Stoltz. Pure Exploration in Multi-Armed Bandits Problems. Proceedings of the 20th International Conference on Algorithmic Learning Theory (ALT’09), 2009.
• [CKLB12] Stephane Caron, Branislav Kveton, Marc Lelarge, and Smriti Bhagat. Leveraging Side Observations in Stochastic Bandits. Uncertainty in Artificial Intelligence, 2012.
• [HK12] Elad Hazan, Satyen Kale. Online Submodular Minimization. Journal of Machine learning Research (JMLR), 13(Oct):2903−2922, 2012.
• [LCLS10] Lihong Li, Wei Chu, John Langford, Robert E Schapire. A Contextual-Bandit Approach to Personalized News Article Recommendation WWW 10, Volume: 173 (2010)
• [S09] Contextual Bandits with Similarity Information Aleksandrs Slivkins Proceedings of the 24th annual Conference On Learning Theory, Issue: June (2009)
• [YM11] J-Y Yu and S. Mannor. Unimodal Bandits. International Conference on Machine Learning (ICML), 2011
This call is posted at http://researchers.lille.inria.fr/~valko/hp/call-phd-2013.php.
For further information please send an email to michal.valko-at-inria.fr as soon as possible.

Michal Valko (equipe SequeL)

CENTRE DE RECHERCHE LILLE – NORD EUROPE
Parc scientifique de la Haute Borne, 40 Avenue Halley, Bât A – Park Plaza
59650 Villeneuve d’Ascq, Office #5 Tel : +33 (0)3 59 57 7801 www.inria.fr
Suivez‐nous sur Twitter twitter.com/inria et YouTube youtube.com/inriachannel

POST-DOC POSITION Telecom ParisTech – Safety Line

POSTDOCTORAL POSITION
possibly leading to a PERMANENT RESEARCH POSITION

Statistician, specialized in risk management

Context: Safety Line offers a permanent contract starting by a one-year post-doc
position co-supervised by the lab of Information and Communication Theory
(LTCI) of Telecom Paristech.

Safety Line is a French innovative company specialising in digital technology for risk
management. It develops statistical algorithms for data mining of data recorded in
flight by airlines (of the magnitude of hundreds of parameters per seconds per flight),
in order to contribute to solving complex problems such as the detection of “weak
signals” in anticipation of disasters. More precisely, from an operational point of view
it is crucial to be capable of identifying “risky sequence”, that is to say a combination
of actions and of phenomena that could lead to damages. In order to enhance the
performance of the analysis, supervised learning-based algorithms which are already
explored should be complemented by unsupervised methods. However, the data are
not well separable, therefore the goal of the research is to develop a general-purpose
detection framework to sort out “meaningful” types of sequences. By “meaningful
type of sequence” we mean a class of time-series which show comparable variations,
in order to point out non-consistent pilot actions for instance.

Activities & Responsibilities: The research topics include but are not limited to:

– Developing statistical and computational methodologies for the analysis of highdimensional
multivariate time-series (flight data or biomedial data), especially for
the risky sequence recognition problems;
– Integrating these methods in a software dedicated to the Risk Management.

From the scientific point of view, the performance of purely unsupervised
methodologies should be compared with semi-supervised learning techniques
integrating risk management expertise.
Deliverables are mainly routines and research reports that detail the performance
comparison between the various techniques used. A particular attention should be
given to the computing performance, in the perspective of an operational
implementation of the algorithms.
The computer developments are performed in C++ or Python, in order to facilitate the
possible embedding of algorithms in the C++ programs that are already in use.
Databases are of type SQL and NoSQL (MongoDB). An access to the computational
resources of Telecom’s lab is granted in order to take advantage of high-end
capabilities. Furthermore, Safety Line is developing its own high-performance
computation capability in collaboration with INRIA and GENCI.

Environment:
The post-doc shall be supervised by Prof. Stéphan Clémençon (Telecom ParisTech)
and Sébastien Travadel (Safety Line).

The LTCI groups all the research activities of Telecom ParisTech (Paris Institute of
Technology), which represents today approximately 160 permanent researchers and
teachers (among which 28 researchers of the National Center for Scientific Research
(CNRS), 270 Ph.D. students as well as 60 post-docs. The Statistics & Applications
Group of the LTCI enjoys a high national and international recognition with editorial
board members in high quality journals (Bernoulli, ESAIM P&S, the Journal of the
Royal Statistical Society, DSP journal) as well as regular participation as program
comity members in the major international conferences(IEEE ICASSP, International
Conference on Machine Learning).

Website:
http://www.ltci.telecom-°©‐paristech.fr/
Safety Line offers an exciting scientific environment to create cutting-edge solutions
for hazard prevention. The team brings together experts in risk management and
researchers in the fields of statistics & human performance. Strong collaborative links
with internationally recognized labs ensure a high-level research program. Indeed,
the company is actively interacting with high-profile scientists in mathematic fields
(statistical & optimization), high performance computation and clinical research on
human performance. The research and development team is a dynamic team of
young professionals dedicated to safety and sustainable development, which reflects
the core values of the firm.
While initially focused on aviation, the activity of the company is developing rapidly
and is now opening on new types of industrial risks. Website: www.safety-line.fr
The successful candidate will work in our Paris-based office (15, rue Jean-Baptiste
Berlier, 75013 Paris). The daily hours are 9 to 5 pm, 5 days a week. 5 weeks of
holidays granted. During the first year (post-doc), working time will be shared as
follows: 4 days/week in Safety Line’s premises, 1 day/week in LTCI’s facilities (37/39,
rue Dareau, 75014 Paris).
The team is multinational and the communication can be in English or in French.

Position Qualifications: Ph.D. in statistics or other relevant, closely related
quantitative field (statistical physics for instance), strong quantitative research
background, statistical and programming proficiency. We are seeking an individual
with a strong background in practical statistical machine learning and computation.

Salary range: 37 to 40 k€ per year + bonus.

Application deadline: 31/03/2013

Benefits: The position offers health coverage, unemployment benefits, pension
contribution and maternity leave.

Application: Applications should include a resumé, brief statements of research
interests, publication records and a link towards the Ph.D. thesis. Applications and
letters should be sent via electronic mail to:
clemenco@telecom-paristech.fr & sebastien.travadel@safety-line.fr