Accueil > A la une

Offre d’emploi / Job offer : postdoc en modélisation bayésienne pour l’écriture avec les yeux au LPNC

publié le , mis à jour le

Le LPNC dispose d’une offre d’emploi de recherches post-doctorales, pour 12 mois, à partir de la rentrée 2013. Les recherches ont pour cadre le projet ANR "Eye On Line", et l’objectif principal est la réalisation d’un logiciel de reconnaissance en-ligne de caractères écrits avec les yeux, basé sur un modèle probabiliste. Les recherches seront effectuées sous la supervision de Julien Diard, en collaboration avec Jean Lorenceau (CRICM, Paris).

Titre : Postdoctoral position : Bayesian modeling of on-line character recognition in an eye-writing software application

Candidates are invited to apply for a 12-month postdoctoral position (to start in September or October, 2013) to study and develop a Bayesian computational model for on-line character recognition. Research will take place in the Laboratory of Psychology and NeuroCognition (LPNC ; CNRS and Grenoble University) in Grenoble, France, under the supervision of Dr. Julien Diard.

The context of this project is the computational study of "eye writing", and software development of character recognition. We use a novel apparatus, based on a particular visual display and the reverse-phi illusory motion, that enables users to generate smooth-pursuit movement at will and in the direction of their choice. Coupled with eye tracking, the system allows participants to "write" cursive letters with their eyes. The application of the system to disabled, motor-impaired patients is central to the project.

Our objectives are two-fold. The first concerns adapting a previous model for cursive character recognition and production (Gilet et al., 2011) to the task of eye writing (Lorenceau, 2012). Our main goal here is to provide rapid character recognition, in an on-line manner, that is to say, as the character is being traced, and robust to the signal characteristic specific to eye writing (e.g., spurious saccades). The second objective is to expand and adapt the model, for instance toward disability assessment, and toward code convergence for easier man-machine interaction (i.e., adapting the system's vocabulary to symbols that are convenient to produce for the patient).

Applicants must recently have obtained a PhD degree in Computational cognitive modeling, artificial intelligence, signal processing or a closely related field. This PhD is required to have been obtained not more than 24 months before the starting date of this postdoc (e.g., starting the postdoc on September, 1st, 2013 if you defended your PhD after September, 1st, 2011). The applicant must not have held any position, previously, in the LPNC. The postdoc will be supported by the EOL (Eye On-Line) project, funded by the French National Research Agency (ANR), in collaboration with Dr. Jean Lorenceau. Gross salary will be 2,500 € / month.

Required skills include software development, signal processing and probabilistic (Bayesian) modeling. The ability to communicate scientific ideas both orally and in writing are essential, while an interest in cognitive science and experimental psychology is desirable. Publications at an excellent level will be expected during the postdoc. Support for administrative procedures, for international candidates, is provided by the ISSO of Grenoble University.

Please send your application to Dr. Julien Diard. Only applications received before 12.00 midday on June, 14, 2013 can be considered (with a tentative decision date on June, 28). You will be required to provide a covering letter, a CV including a publication list, and names and contact information of two references with a brief description of your relationship to each reference. Applications can be written in French or English. Working language, also, will either be French or English. Informal enquiries before submitting a full application are welcome.

Contact Person : Dr. Julien Diard
Contact Phone : (+33) 4 76 82 58 93
Closing Date : 14 June 2013
Contact Email :

Online resources for more information :