Julien CORNEBISE, PhD, Postdoctoral Research Associate
|
Department of Statistical Science
University College London
Gower Street, London WC1E 6BT
United Kingdom
Access
Office: 103
Phone: +44 20 76 79 12 23
E-mail: julien AT cornebise.com
Public key ID: 1024D/2D2BC3F8
Key fingerprint: 4EC8 596D 9006 1AF0 0DF3 4866 2984 7780 2D2B C3F8
|
Current status
I am currently a Postdoctoral Research Associate
in the Department of Statistical Science of University College London (London, UK), with Pr. Mark Girolami since mid-December 2010.
Before that, in 2009 and 2010 I did a two-years joint postdoct at University of British Columbia (Vancouver, Canada) and Statistical and Applied Mathematical Sciences Institute (SAMSI) (North Carolina, USA), with Pr. Arnaud Doucet. To know more about my path until there, you are welcome to
have a look at my (somewhat outdated) complete
CV -- also available in French.
I am a researcher in the field of Applied Mathematics, working at the
frontier of Computational (Bayesian) Statistics and Applied Probability. My current research interests lie in the following fields:
- Riemannian Markov Chain Monte Carlo (MCMC) methods, using concepts of information geometry to sample from intricate distributions arising typically in nonlinear dynamic systems -- with applications in System Biology (e.g. signalling pathways).
- Sequential Monte Carlo (SMC) methods, especially theory and methods, with still a strong interest on concrete applications. This includes non-exhaustively SMC filtering, SMC samplers, ABC-SMC.
- Adaptive Monte-Carlo methods, i.e. self-tuning of these algorithms with respect to the data so as to maximize their efficiency.
In the past, I have also been involved in :
- Stochastic optimization, such as SAEM, MCEM, and its variants, extremely useful in the former two points.
- Nonlinear design of experiments.
For more about all this, you are invited to check my publications list.
Ph.D. Thesis
News on April 4th, 2011: I have the great
honor of being one of the two finalists of the Savage Award 2010 by
the International Society for Bayesian
Analysis (ISBA), category Theory and Methods. The final result will be announced
at the Joint Statistical Meeting 2011 early August.
I deposited my manuscript on March 10th, 2009, and defended my Ph.D. on June 25th, 2009.
- Its title is Adaptive Sequential Monte Carlo Methods and you can download the manuscript.
- My advisors were:
- Pr. Eric Moulines, from
STA team of Laboratoire Traitement du Signal et des
Images (TSI), at Telecom ParisTech, and
- Pr. Paul Deheuvels, from Laboratoire de Statistique Théorique et Appliquée (LSTA), in University Pierre and Marie Curie - Paris 6,
not forgetting my informal co-advisor and precious collaborator Dr. Jimmy Olsson (Lund University, Sweden).
- My defense's jury was:
- Pr. Paul DEHEUVELS (Advisor)
- Pr. Eric MOULINES (Advisor)
- Dr. Fabien CAMPILLO (Reviewer)
- Pr. Paul FEARNHEAD (Reviewer)
- Pr. Christophe ANDRIEU (Examiner)
- Pr. Gérard BIAU (President)
- Pr. Arnaud DOUCET (Examiner)
- Pr. Christian ROBERT (Examiner)
- I was funded by a french MNRT ministerial grant from October 2005 to October 2008, then by a CNRS contract from October 2008 to December 2008.
- Additionally, I was a moniteur (T.A.) in University Paris Diderot - Paris 7 from October 2005 to October 2008, as well as occasional chargé de cours (lecturer) at Telecom ParisTech in 2006--2007 and 2007--2008. More on this...
Additional scientific interests
I am most curious about, or even addicted to, everything involving mathematics or computer science. Although I had to realize years ago that understanding all of the maths is way out of reach, I am eager to discover (even elementarily) as much as I can. Especially - but in no way exclusively - in:
- Algorithmic - aftermath of participation to the wonderful ACM International Collegial Programming Contest,
- Combinatorial optimization-- especially by means of stochastic algorithms,
- Problem-solving,
- or any other domain that can twist your mind to despair while you do not find the answer, and
finally seems horribly obvious once reached.
Next place to meet me
Want to chat ? Welcome ! You can either simply e-mail or phone me, so as to meet in my office or in café
- parisian cafés are excellent places to do maths, US and Canada did well too, now trying Londonian ones.
Last modified: 12/23/2011 06:20 PM