Institut des Systèmes Intelligents
et de Robotique

Partenariats

UPMC

CNRS

INSERM

Tremplin CARNOT Interfaces

Labex SMART

Rechercher

A voir également

Short bio

liénard Jean
Title : PhD Student
No longer in the unit

 

 

I am a PhD student since Dec. 2009, under the supervision of Benoît Girard.

Subject: Artificial evolution of neuro-mimetic models of action selection

Keywords: computational neurosciences, evolutionary algorithms, action selection, basal ganglia, medial reticular formation

In a few words...

The use of evolutionary algorithms in computational neurosciences has been confined to scattered attempts. Most of the time, they consist in a simple adjustment of parameters at the end of the model conception, when every aspect of the architectures are fixed, and with basic genetic algorithm. The aim of my work is the use of modern evolutionary algorithms that go beyond simple optimization.

The basal ganglia are a complicated set of interconnected subcortical structures that process motor, cognitive and motivational (limbic) cortical information. They can be viewed as a generic mechanism of selection. At the moment, too few computational model can claim to unify the large amount of anatomical knowledge on the basal ganglia while achieving to fit the known electrophysiological data. I believe that innovative use of evolutionary algorithm can overcome the rise in complexity of elaborating basal ganglia models, and can thus help to understand the functional meaning of the anatomical and electrophysiological knowledge.

The medial reticular formation, located in the midbrain, seems to be constituted by the replications of elementary networks along a particular topology. Only a few works tried to model them. In a prospective way, the use of evolutionary algorithms to create a plausible computational modeling could provide insights on their functionality and anatomy.