CORTEX Research team
- Leader : Bernard Girau
- Research center(s) : CRI Nancy - Grand Est
- Field : Computational Sciences for Biology, Medicine and the Environment
- Theme : Computational Medicine and Neurosciences
- Partner(s) : CNRS,Université de Lorraine
- Collaborator(s) : CNRS, INRIA
The goal of the CORTEX project-team is to study the properties and capacities of neural computation, seen as distributed, numerical and adaptative information processing. More precisely, we intend to show that such processing can allow for the realisation of intelligent systems, able to extract knowledge from data and to manipulate that knowledge to solve problems. These capabilities are obtained by defining neuronal connectionist models developped through two sources of inspiration, namely computational neurosciences and machine learning.
Firstly, at a global level, we are developping behavioral neuronal models, to allow for the navigation of an autonomous robot. With this aim, we are implementing and interfacing models of associative cortex (sensorimotor coordination), prefrontal cortex (temporal organisation of behavior) and hippocampus (localisation). Secondly, at a lower level, we are studying elementary neuronal models, based on the binary functioning of the neuron (spike emission). This work corresponds to studying temporal phenomena within assemblies of such neurons. Thirdly, at an intermediate level, we are working on classical continuous neuronal models and are studying their adaptation to problems of database exploration, with a view to neurosymbolic integration (knowledge extraction and integration). These computer science works are mainly implemented on classical computers, but we are also exploring other ways such as parallel machines, autonomous robots and more generally specialised circuits for embedded systems.
International and industrial relations
- European projects: MirrorBot, Scholnet; networks of excellence Neuronet and Neurocolt.
- Contracts: Usinor, France Télécom, Thomson, ACI Cognitique.
- University relations: Universities of Geneva, Amsterdam, Boulder, Missouri, Sunderland, Plymouth, Taiwan, Montréal; Universities of Paris VI, Strasbourg, Lyon, Inserm.
Keywords: Connectionism Neural networks Action/perception interaction Reinforcement learning Neuro-symbolic integration Artificial intelligence Cognitive phychology Knowledge representation Signal interpretatio