Call for Papers: Foundations and New Trends of PAC Bayesian Learning

Workshop:

Foundations and New Trends of PAC Bayesian Learning

University College London, UK

22 – 23 March 2010

http://www.cs.ucl.ac.uk/staff/rmartin/pacbayes/

CALL FOR PAPERS
Deadline: Friday, 12th February 2010

PAC-Bayes theory is a framework for deriving some of the tightest generalization bounds available. Many well established learning algorithms can be justified in the PAC-Bayes framework and even improved. PAC-Bayes bounds were originally applicable to classification, but over the last few years the theory has been extended to regression, density estimation, and problems with non iid data. The theory is well established within a small group of the statistical learning community, and has now matured to a level where it is relevant to a wider audience. The workshop will include tutorials on the foundations of the theory as well as recent findings through peer reviewed presentations.

Workshop topics

PAC Bayes theory or applications. In particular: application to:

* regression
* density estimation
* hypothesis testing
* structured density estimation
* non-iid data
* sequential data

The Invited Speakers include:

Olivier Catoni
CNRS U.M.R. 8553

David McAllester
Toyota Technological Institute at Chicago

Matthias Seeger
Saarland University and Max Planck Institute for Informatics

Organisers: Jean-Yves Audibert, Matthew Higgs, Steffen Grünewälder, François Laviolette and John Shawe-Taylor

Contact:
Steffen Grünewälder
steffen(at)cs.ucl.ac.uk

PhD (and postdoc) opportunities in Bristol

There are currently a number of opportunities if you are interested in research
at the Intelligent Systems Laboratory of the University of Bristol –
both at the postdoc level and PhD student level,
some for UK students and other for overseas. Deadlines are tight, so please read carefully.

Rsearch topics in the ISL range from statistical learning to web mining, including large scale
data mining and bioinformatics. My own group is mostly focussed on massive scale
pattern analysis, web mining, intelligent systems design, theoretical models of machine learning.

The University of Bristol is offering postgraduate scholarships (competitive)
(Home/EU and Overseas). The deadline for applications is 1st March 2010 for *both*.

http://www.bristol.ac.uk/studentfunding/overseas_pg/overseas_schols.html
http://www.bris.ac.uk/studentfunding/home_pg/schols.html

Separately, if you are interested in postdoc research in the areas listed above
Bristol could be the ideal place for this postdoc scheme:
fellowships starting in January 2011 – information is available at www.newtonfellowships.org . The deadline for applications is 8 February 2010.

Call for Papers: ACL 2010 Workshop on Applications of Tree Automata in Natural, Language Processing (ATANLP)

Call for Papers: ACL 2010 Workshop on Applications of Tree Automata in Natural
Language Processing (ATANLP)

July 16, 2010
Uppsala, Sweden

http://stp.lingfil.uu.se/atanlp/

1. Workshop Description

The theory of tree automata has an increasing number of applications in natural
language processing; examples for this can be found in work on topics as diverse as
grammar formalisms, computational semantics, language generation, and machine
translation. Here, the term tree automaton is to be understood in a generic sense,
including all types of formal devices that specify, generate or transform tree
languages or, more generally, tree series.

The goals of this workshop are to provide a dedicated venue for the presentation of
work that relates the theory of tree automata to natural language processing, and to
create a forum where researchers from the two areas can meet and exchange ideas.
Specifically, the workshop aims at raising the awareness for theoretical results
useful for applications in natural language processing, and at identifying open
theoretical problems raised by such applications.

2. Topics of Interest

Topics of interest for the workshop are all topics related to or motivated by the
application of tree automata in natural language processing. These include but are
not limited to the following:

* representations of languages by tree automata and tree grammars;
* tree transducers, synchronous grammars, and related devices and their application
to, e.g., machine translation;
* tree logics and their application to, e.g., natural language syntax and semantics;
* weighted extensions of the aforementioned;
* algorithms related to the implementation of tree automata, such as algorithms for
matching, accepting, and parsing;
* learning and training algorithms for tree automata;
* tree automata-based query languages for, e.g., treebanks and parallel syntactic
corpora;
* relations between tree languages and string languages with motivation in NLP;
* case studies concerning the application of tree automata techniques in natural
language processing.

3. Invited Speaker

Kevin Knight (ISI/University of Southern California, USA)

4. Submission Information

The workshop invites submission of two kinds of contributions: full papers and
proposals for so-called quickfire presentations.

4.1 Full Papers

Full papers should report original and unpublished research on topics of interest
for the workshop. Accepted papers are expected to be presented at the workshop, and
will be published in the workshop proceedings. They should emphasize obtained
results rather than intended work, and should indicate clearly the state of
completion of the reported results. A paper accepted for presentation at the
workshop must not be presented or have been presented at any other meeting with
publicly available proceedings. If essentially identical papers are submitted to
other conferences or workshops as well, this fact must be indicated at submission.

Reviewing will be double-blind, and all papers will receive at least three
independent reviews. Submissions will be assessed with respect to appropriateness,
clarity, soundness/correctness, meaningful comparison, originality/innovativeness,
and impact of ideas or results.

The maximum length of a submitted paper is eight (8) pages of content, excluding
references. If necessary, authors may add an appendix containing proofs and the
like, but the paper should be accessible without reading the appendix. The final
manuscript is limited to eight (8) pages of content and nine (9) pages in total.

As reviewing will be double-blind, the paper should not include the authors’ names
and affiliations. Furthermore, self-references that reveal the author’s identity,
anonymous citations and acknowledgements should be avoided.

4.2 Proposals for Quickfire Presentations

Individual researchers and research groups working on topics of interest for the
workshop are invited to present their work to the workshop audience in the form of a
so-called quickfire presentation of about 10 min each. The idea behind these
presentations is to provide a means for triggering discussions and an exchange of
ideas. Based on the interests of a research group or an individual researcher, the
typical quickfire presentation will point out potential relations between the theory
of tree automata and natural language processing, thus indicating ideas and
opportunities for future collaboration rather than presenting specific results
obtained. In particular, we welcome quickfire presentations by researchers in
natural language processing who wish to enter into a scientific discussion with
researchers in tree automata theory, or vice versa.

Quickfire presentations will be selected based on informal proposals that briefly
describe the planned content of the presentation and indicate why it is supposed to
be of interest for the other attendees of the workshop. The maximum length of a
submitted proposal is one (1) page. Proposals for quickfire presentations related to
submitted full papers should carefully point out the difference between the two.

4.3 General Information

All submissions must be electronic in PDF and must be formatted using the ACL 2010
style files, which are available at the following address:

http://www.acl2010.org/authors.html

Contributions should be submitted via the submission site:

https://www.softconf.com/acl2010/ATANLP/

The page limits have to be be strictly observed; submissions exceeding them will not
be considered. Final decisions on the program will be made by the Programme
Committee.

The submission deadline is 23:59 CET on 5 April 2010.

5. Important Dates

Submission deadline: April 5, 2010
Notification of acceptance: May 6, 2010
Camera-ready versions due: May 16, 2010
Workshop: July 16, 2010, following ACL 2010

6. Workshop Chairs

Frank Drewes,
Department of Computer Science, Umeå University, Sweden

Marco Kuhlmann,
Department of Linguistics and Philology, Uppsala University, Sweden

The workshop chairs can be contacted at the joint email address atanlp2010 AT
fastmail DOT net.

7. Programme Committee

Parosh Aziz Abdulla (Uppsala University, Sweden)
Leonor Becerra-Bonache (Yale University, USA)
Chris Callison-Burch (Johns Hopkins University, USA)
David Chiang (ISI/University of Southern California, USA)
Loek Cleophas (University of Pretoria, South Africa)
Trevor Cohn (University of Edinburgh, UK)
François Denis (Université de Provence, France)
Thomas Hanneforth (Universität Potsdam, Germany)
Johanna Högberg (Umeå University, Sweden)
Liang Huang (ISI/University of Southern California, USA)
Stephan Kepser (codecentric GmbH, Germany)
Alexander Koller (Saarland University, Germany)
Andreas Maletti (Universitat Rovira i Virgili, Spain)
Sebastian Maneth (NICTA, Australia)
Jonathan May (ISI/University of Southern California, USA)
Brink van der Merve (University of Stellenbosch, South Africa)
Mark-Jan Nederhof (University of St Andrews, UK)
Joachim Niehren (INRIA, France)
Kai Salomaa (Queen’s University, Canada)
Anoop Sarkar (Simon Fraser University, Canada)
Giorgio Satta (University of Padua, Italy)
Stuart Shieber (Harvard University, USA)
Magnus Steinby (University of Turku, Finland)
Marc Tommasi (INRIA, France)
Heiko Vogler (Technische Universität Dresden, Germany)

ECML PKDD 2010 – Call for papers

ECML PKDD 2010 – Call For Papers
The European Conference on Machine Learning and
Principles and Practice of Knowledge Discovery in Databases
September 20-24, 2010
Barcelona, Spain
http://www.ecmlpkdd2010.org/

Key Dates:
———-
* Abstract due: April 23, 2010
* Papers due: April 30, 2010
* Notification: June 21, 2010

ECML PKDD is among the premier scientific forums in machine learning
and data mining: this call invites you to submit your latest results
in these areas.

All papers presenting new and insightful results in machine learning
and data mining are welcome, specifically papers addressing
(non exhaustive list):

* new settings and goals (from interactive learning to privacy);
* new hypothesis spaces and algorithms (deep, hybrid, combinatorial,
condensed, infinite, and beyond);
* new applications (robotics, games, environmental, social, text);
* computational issues (hierarchical, high performance, decentralized);
* negative results: where do the frontiers of intractability lie?

Selection criteria include soundness and reproducibility, originality,
boldness with respect to theory or applications.

Submissions:
————
Authors should not submit any paper which is under review or which has been
accepted for publication in a journal or another conference; neither should
they submit their papers elsewhere during the review period of ECML PKDD 2010.

Papers must be written in English and formatted according to the Springer LNCS
guidelines: http://www.springer.de/comp/lncs/authors.html
The maximum length is 16 pages.

Proceedings and Journals:
————————-
The conference proceedings will be published by Springer, Lecture Notes
in AI series. Further, following last years tradition, a selection
of the submitted papers will be published directly in two special issues,
of Machine Learning Journal and Data Mining and Knowledge Discovery
respectively.
See also Springer ECML PKDD Community Portal:
http://www.springer.com/computer/lncs/ecmlpkdd

Follow ECML PKDD news on Twitter:
http://twitter.com/ecmlpkdd10

PhD Studentship in Computational Ecology

Ecological interaction networks: Application to ongoing field work in Tanzania

Ecosystems consist of a complex web of interactions: among species, and between species and their environment. Understanding these interactions is important for predicting how factors – such as biological or chemical control agents, grazing pressure, or climate change – may affect ecosystem function. However, unravelling such complex networks of interactions stretches the boundaries of current research; new methods are needed to handle the complexity of real ecosystems.

This studentship explores the use of a methodology new to ecological analysis, Bayesian networks, for revealing ecological interaction networks. You will develop Bayesian network algorithms for handling ecological data and incorporating spatial information. Methods developed will be applied to real ecological data, including data collected in an ongoing field project in Tanzania. There is opportunity for a field season in Tanzania, where you would perform targeted manipulation experiments based on your models (e.g., targeted removal and addition of species within caged areas).

This project will be based in Dr Anne Smith’s lab at St Andrews and be in collaboration with Dr Colin Beale (York University, currently based in Tanzania) and Dr Dirk Husmeier (Biomathematics & Statistics Scotland). In this studentship, you will gain both computational and field skills.

For more details, please contact anne.smith (at) st-andrews.ac.uk and/or visit: http://biology.st-andrews.ac.uk/vannesmithlab/

Funding by NERC: UK residents (stipend and fees) or EU citizens (fees only)

A PDF of the above advertisement is downloadable from:

http://biology.st-andrews.ac.uk/vannesmithlab/NERC2010.pdf

Conference on Bayesian Nonparametric Statistical Methods at Santa Cruz, CA, Aug 2010

CBMS Regional Conference – Bayesian Nonparametric Statistical Methods: Theory and Applications. Santa Cruz, California, August 16-20, 2010.

Bayesian nonparametric (BNP) methods combine the advantages of Bayesian modeling (e.g., ability to incorporate prior information, full and exact inference, ready extensions to hierarchical settings) with the appeal of nonparametric inference. In particular, they provide data-driven, albeit model-based, inference and, importantly, more reliable predictions than parametric models.

Theoretical research on NPB methods and their applications has grown dramatically in the last fifteen years. This has produced a massive body of scattered literature, which can be daunting for newcomers and hard to follow even for specialists. This CBMS conference, to be held between August 16th and August 20th, 2010 at the campus of the University of California, Santa Cruz, aims at providing a comprehensive introduction to the field for new researchers, and in particular graduate students, postdocs and junior researchers.

The main lecturer for the conference will be Dr. Peter Muller, who is Robert R. Herring Distinguished Professor in Clinical Research in the Department of Biostatistics at the M.D. Anderson Cancer Center, Houston, TX. In addition to the ten lectures delivered by Dr. Muller, four invited speakers will deliver complementary two-hour lectures. These invited speakers include Michael Jordan (University of California, Berkeley), Peter Hoff (University of Washington), Wesley Johnson (University of California, Irvine) and Tim Hanson (University of Minnesota). The local organizers are Abel Rodriguez and Athanasios Kottas.

More information can be found at the conference website https://www.ams.ucsc.edu/CBMS-NPBayes or through email at CBMS-NPB@ams.ucsc.edu

Newton International Fellowship Scheme – Deadline 8 February 2010

Dear academic,

The next round of the Newton International Fellowship scheme is currently open and closes on 8 February 2010. The scheme, run by The British Academy, The Royal Academy of Engineering and the Royal Society, aims to attract the world’s best postdoctoral researchers to the UK for two years.

The Fellowships cover the broad range of natural and social sciences, engineering and the humanities and are open to early-stage postdoctoral researchers who do not hold UK citizenship and are working outside the UK. Fifty Fellowships are available per round and successful candidates will receive an annual subsistence of £24,000, up to £8,000 for research expenses, and a one-off payment of up to £2,000 for relocation.

Newton Fellows may be eligible for follow-on funding of up to £6,000 per year for ten years, to help develop lasting international networks with the UK.

Applications are invited for Fellowships starting in January 2011 and further information is available atwww.newtonfellowships.org . The deadline for applications is 8 February 2010. Please bring this scheme to the attention of any colleagues who may be interested.

Thank you for your help in promoting this important scheme.

Newton Operations Team

Postdoctoral Position, CRAN, France

Our laboratory proposes a postdoctoral position in the field of fault detection and diagnosis (FDI) based on kernel PCA (principal components analysis). We are looking for a candidate with competences in data analysis and machine learning.

More information available at http://www.pascal-network.org/vacancies/postdoc_eng_abs.pdf

Post-doc Position – Intelligent Systems Laboratory, Bristol

We are currently inviting applications for a post-doc position in the Intelligent Systems Laboratory of Bristol.

Topic: statistical techniques for informative pattern mining in complex and structured data, and applications.
Applicants with a theoretical interest in this subject, as well as applicants with a specific interest in applications to content-based music information retrieval, are particularly welcome to apply.

The closing date for applications is 9am 1 March 2010, and the interview will be in early March 2010.
The starting date is flexible and to be agreed, but will be no later than 1 October 2010.
The duration of the post is up to 30 months.

Further details of the post and application procedure are available from:
http://www.jobs.ac.uk/job/AAM962/research-assistant-/

COLT 2010: Call for Open Problems

COLT 2010 will include a session devoted to the presentation of open problems.
A description of these problems will also appear in the COLT proceedings.

The write-up of an open problem should include the following:

1. A clear, self-contained description of an open problem
2. Motivation for the study of this problem
3. The current state of understanding for this problem, including known
partial solutions and citations of related published work

We especially encourage people to propose descriptions of new interesting
research directions in areas that are currently outside the scope of COLT,
such as bioinformatics, privacy and security, and vision, to name a few.
Ideally, your open problems or research directions should include well-defined
mathematical questions, nontrivial, and explainable without requiring too
much specialized background knowledge in a 5-10 minutes talk. Monetary
rewards for solving an open problem are encouraged but not required.

Format and submission: The open problems should be 1-2 pages long in the
COLT proceedings format.

Please submit them electronically to open@colt2010.org with subject line
“open problem for Colt2010”. The submissions, in pdf or ps, should be
attached to the email.

Deadline: March 13, 2010.