Connectionists: Postdoc position on Learning and Emotions

[Frederic Alexandre]

A postdoc position is available in the CORTEX group, INRIA-Nancy, 
France, in the domain of Learning and Emotions to begin by end of 2008. 
Candidates should apply on-line (see address below) but can also 
directly contact me for any question.

alexandre

http://www.inria.fr/travailler/mrted/en/postdoc/details.html?id=PNGFK026203F3VBQB6G68LOE1&LOV5=4508&LOV2=4490&LOV6=4513&LG=EN&Resultsperpage=20&nPostingID=1896&nPostingTargetID=4868&option=52&sort=DESC&nDepartmentID=19 


Learning and Emotions
Position type: Post-doctoral Fellow
Functional area: Nancy
Research theme: Cognitive systems
Project: CORTEX


Environnement
http://cortex.loria.fr/

Missions
In the domain of machine learning, learning is traditionally considered 
as an optimisation process, where one tries to appromimate as accurately 
and as quickly as possible a function to get the best performances with 
regard to a global criterion.
As studies from biology and psychology about the role of emotions in 
learning are considered, it appears that learning is not a linear 
neither global process : each new example or situation does not trigger 
the same level of learning but is modulated by emotions. For example, it 
is sufficient to be burned one time to learn that fire is dangerous. 
Also, simple innate or learned emotional reactions can be used as a 
scaffolding to realize more complex learning. For example being 
attracted by novelty is interesting to quickly explore a scene and can 
be a solution to the curse of dimensionality.
In the framework of the embodiment of cognition and more particularly of 
autonomous robotics, it is important to modulate learning, depending on 
the situations, particularly to manage critical situations and to 
incrementally learn in highly dimensional spaces, as encountered in real 
environments.
The goal of this postdoctoral position is to help us think to the 
emotional solution to develop new methodologies of learning, to find in 
the neuroscience domains structural and functional hints about the role 
of emotions in learning and to study how these hints can be introduced 
in neuronal modules compatible with those developped in our team.

Activités
Step 1 : bibliographical studies on two domains : (i) description in 
different domains of neuroscience (physiology, cognitive neuroscience, 
behavioral biology) of neuronal expression (central and peripheric) of 
emotions and their impact on the rational brain ; (ii) familiarization 
with the formalism of neuronal computations developped in the team 
(neural fields) and with sensorimotor models previously developped.
Step 2 : study and experimentation of bio-inspired models previously 
developped in this perspective and particularly pavlovian conditioning 
models (ex : Schmajuk), amygdala models (ex : Moren), models linking 
basal ganglia with the prefrontal lobe (ex : Doya, Girard). The software 
developments will be made with the formalism of neuronal computation 
developped in the team (neural fields).
Step 3 : development of robotic experiments : choice of a scenario of 
incremental learning ; determination of emotional stimuli and of their 
embodiment ; realisation of a protocole of interaction of the robot with 
the environment including a series of learning stages from repondant 
conditioning to instrumental conditioning.
Campaign of tests to determine the performances and characteristics of 
such a learning and comparison with simple sensorimotor learning.

Compétences et Profil
First of all, the applicant should have an experience and education in 
cognitive neuroscience, particularly in the emotional domain. An 
experience in bio-inspired neuronal modeling is also an important point. 
Abilities in computer science should be mentionned. A knowledge and 
practice in neural fields would be appreciated.

Duration of the contract : 12 to 24 months
-- 

Frederic ALEXANDRE

http://www.loria.fr/~falex/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.srv.cs.cmu.edu/mailman/private/connectionists/attachments/20080327/a81a4c91/attachment.html