UCL, Lecturership in Statistics

Lecturer in Statistics, UCL Department of Statistical Science

Applications are invited for a Lectureship available in the Department of Statistical Science, University College London from 1st September 2009.

Applicants should have research strengths in any branch of Probability and Statistics and an aptitude for teaching. They should also have, or expect to have in the near future, a PhD in Statistics or a related discipline. The successful applicant will be expected to carry out original research and to contribute to undergraduate and postgraduate teaching in Statistical Science.

An appointment will be made within the salary ranges £32,458 per annum
-
£35,469 per annum (Lecturer A) or £36,532 per annum - £43,622 per annum (Lecturer B), dependent upon skills and experience. London Allowance of £2,781 per annum will be paid in addition.

Further particulars, including the job description and person specification, can be viewed at http://www.ucl.ac.uk/Stats/department/jobs.html or may be obtained from Marion Ware, Departmental Administrator, telephone 020 7679 1872. Letters of application, including a CV and names of three referees, should be sent to:

Head of Department, Department of Statistical Science,
University College London, Gower Street, London WC1E 6BT. Tel: 02076791872,
Fax: 020 7383 4703, email: hod@stats.ucl.ac.uk.

UCL Taking Action for Equality.

The closing date for applications is Friday, 6th March 2009.

Vacancy PostDoc, Statistical NLP

The Institute for Logic, Language and Computation
http://www.illc.uva.nl

Postdoctoral Researcher

1.0 fte (38h per week)

For both internal and external candidates

Summary

• Position: POSTDOC researcher
• Duration: 2 years (full time, 38hrs per week)
  
• Salary (gross per month): minimum Euro 2379 and maximum Euro 3755 in the first year.
• The Collective Employment Agreement of the Dutch Universities is applicable.
• Last date for application: February 15, 2009.
  

Project

Priors for the Estimation of Probabilistic Grammars from Incomplete Natural Language Data
VIDI project Sima'an [2007-2011]

Abstract: The project aims at exploring the utility of available prior knowledge in specifying the model (hypothesis space) and the constraints on the objective estimation function for unsupervised estimation of probabilistic grammars for Statistical Machine Translation and Statistical Parsing.
This project addresses problems of unsupervised estimation in the face of different kinds of incomplete data (parallel and monolingual corpora):

1. A raw text corpus,
2. A partially annotated corpus possibly with negative examples, and
3. A raw in-domain corpus and an out-of-domain annotated corpus.
   

Corpus annotation is e.g., word-alignment, syntactic or semantic structure.

Example topics studied within the project (see project related publications):

• Statistical machine translation with hierarchical/syntactic structure, e.g., statistical estimation [Mylonakis and Sima'an 2007 and 2008], lexical syntactic structrure [Hassan, Way and Sima'an 2007 and 2008]
  
• Statistical machine translation with morphological structure -- Work in progress 2009,
  
• Inducing hidden compositional structure for machine translation (possibly using syntactic structure) -- Work in progress 2009.
  

Project team: Markos Mylonakis, Tejaswini Deoskar, and Khalil Sima'an (project leader).
Affiliated researchers: Reut Tsarfaty.
External collaborators: Hany Hassan , Barbara Plank and Andy Way.


Requirements

We seek a cooperative yet independent researcher who is passionate about empirical research, statistical methods and NLP/CL.
The following are the main requirements:

• obtained a Ph.D. recently (or will obtain it in the very near future, in which case testemoney from the PhD thesis Supervisor is needed)
• a strong background, experience in and affinity with the largest subset of the following areas:

1. Computational Linguistics/Natural Language Processing working with statistical methods
2. Machine Learning Applied to NLP with good knowledge of unsupervised statistical methods.
   
3. Statistical Machine Translation.
4. Unsupervised statistical methods for statistical Parsing.

Appointment

The appointment will be on a temporary basis for a maximum period of 2 years. Based on a full-time appointment (38 hours per week) the gross monthly salary will range from € 2379,- to, at the most, € 3755,- , in the first year depending on expertise and experience.

The Collective Employment Agreement of the Dutch Universities is applicable.

Application
To Dr Khalil Sima'an by email (k.simaan@uva.nl with subject heading: VIDI-POSTDOC).
The following documents (in pdf):

1. Letter expressing interest
   
2. Curriculum Vitae
3. List of publications (possibly including a downloadable pdf of your thesis)
   
4. Three letters of reference sent directly by the referees to Dr Khalil Sima'an
   
5. Optional: Research plan (1 page)

Last date for application: February 15, 2009.

For inquiries contact

Dr Khalil Sima'an.
Language and Computation
Computer Science
University of Amsterdam

Research Fellow, Statistics/Machine Learning at University of Edinburgh

Research Fellow (Statistician, Machine Learning Specialist)
£29,704 - £35,469

A research post is available immediately at the University of Edinburgh for a Bayesian statistician/computer scientist to work on developing statistical methods for genetic epidemiology. The object is to develop methods for modelling the joint effects of genotype and environment and inferring causal relationships, using tools such as Markov chain Monte Carlo simulation, variational Bayes approximations, and Dirichlet process models. The successful applicant is likely to learn about and make use of the recently released Infer.NET (http://research.microsoft.com/infernet), as well as developing other techniques.

The post would suit an individual at post-doctoral level with experience of using and developing Bayesian probabilistic methods for modelling and inference. Previous experience in statistical genetics or bioinformatics is useful, but is far from an essential requirement: experience in novel
development and application of probabilistic machine learning and statistical methods in other fields could be just as valuable. The post is funded by the Medical Research Council for three years with possibility of extension should further funding be obtained.

Closing date: 9 January 2009.

See http://www.jobs.ac.uk/jobs/RO886 /Research_Fellow_Statistician_Machine_Learning_Specialist/

Interested applicants can contact Paul McKeigue (paul.mckeigue@ed.ac.uk) or Amos Storkey (a.storkey@ed.ac.uk) for further details.

3rd SIPTA School on Imprecise Probabilities

3rd SIPTA School on Imprecise Probabilities: 1st announcement

[All our apologies for cross-posting.]

Third school of the Society for Imprecise Probability: Theories and Applications (SIPTA), July 2-8, 2008, Montpellier, France.

Dear colleagues,

The Society for Imprecise Probability: Theories and Applications (SIPTA) organizes the 3rd edition of its school in Montpellier, France, during 2-8 July 2008, the local organization being handled by the Laboratoire d'Informatique de Robotique et de Microelectronique (LIRMM) of the university of Montpellier. The city of Montpellier is located in the south of France, on the Mediterranean coast, and the school will be held at the Centre Regional de Documentation Pedagogique, which is located in the very centre of Montpellier.

The aim of SIPTA schools is to introduce interested students and researchers with the basics of imprecise probability topics, both theoretical and applied. Some of the best specialists in different aspects of imprecise probabilities lecture, during one week time, on the main concepts and techniques associated to their area of expertise, in a friendly environment favouring interactions between participants.

Topics & Lecturers

Imprecise probability is used as a generic term to cover all mathematical or statistical models which measure chance or uncertainty without sharp numerical probabilities. Imprecise probability models are needed in inference problems where the relevant information is scarce, vague or conflicting, and in decision problems where preferences may also be incomplete. The topics covered this year will include:

Coherent lower previsions (Enrique Miranda, Spain, and Gert De Cooman, Belgium)
Predictive inference: from Bayesian inference to Imprecise Probability (Jean-Marc Bernard, France)
Imprecise immediate predictions (Gert De Cooman, Belgium)
Independence concepts in Imprecise Probability (Fabio Cozman, Brazil)
Possibility theory (Didier Dubois, France)
Algorithms & approximation methods for Imprecise Probability (Fabio Cozman, Brazil)
Game-theoretic probability and its link with Imprecise Probability (Glenn Shafer, USA)

The final program should be announced shortly on the summer school website.

Pre-registration

Since the number of participants is limited, we recommend you to inform us of your intention to participate to the school as soon as possible, by pre-registering by means of a simple reply to this email, indicating your name, email, status and affiliation, or by filling the pre-registration form at the school website.

More information is available at the school website:

http://www.lirmm.fr/SIPTASchool08/

We are looking forward to meeting you in Montpellier. We also welcome you to circulate this announcement around you.


Best regards,

Jean-Marc Bernard
Kevin Loquin
(for the scientific and organizing committees)