Description du projet
Des spécialistes en chimie et en science des matériaux se joignent à la quête des circuits neuromorphiques
Le cerveau humain a longtemps servi de modèle au monde de l’informatique, qui a essayé de dupliquer ses fonctions grâce à des réseaux basés sur des matériaux en silicium. Plus récemment, l’accent a été mis sur des solutions de substitution au silicium qui imiteraient l’apprentissage des réseaux neuronaux, offrant vitesse, souplesse et fiabilité sans que leur coût ne soit prohibitif. Le projet MANIC formera 15 chercheurs en début de carrière à l’expérimentation de nouveaux matériaux, faisant appel à des chimistes et à des spécialistes des matériaux pour travailler dans le cadre des recherches sur les circuits neuromorphiques. Ils travailleront sur des matériaux qui accéléreront les progrès technologiques visant à mettre au point des plateformes informatiques fonctionnant de manière efficace et flexible sans avoir besoin de beaucoup d’énergie, à la manière du cerveau humain.
Objectif
Large efforts are invested into developing computing platforms that will be able to emulate the low power consumption, flexibility of connectivity or programming efficiency of the human brain. The most common approach so far is based on a feedback loop that includes neuroscientists, computer scientists and circuit engineers. Recent successes in this direction motivate the scientific community to start working on the next big challenge: using materials that emulate neural networks. For that, new players are needed: material scientists, who look into alternatives to silicon in order to develop basic device units, more fitting to the needs of cognitive-type processing than current transistors. We notice that recent progress in chemistry and materials sciences (atomically controlled materials) and nanotechnology (diversity of tools to probe the nanometer scale) brings exciting possibilities for novel approaches in the area of neuromorphic computing. Clearly, the type of materials, physical responses and spatial dimensions considered in the design of neuromorphic systems will crucially determine their utilization, properties and cost, and consequently their societal and economic impact. Therefore, it is urgent that chemists and materials scientists also join forces in the development of the future neuromorphic computer. MANIC aims to offer complementary expertise to current approaches by recruiting fifteen Early Stage Researchers (ESRs) and providing them with the best possible research, academic and professional training, to prepare them for the challenge of developing advanced materials with memory, plasticity and self-organization that will perform better than the current solutions to emulate neural networks and, eventually, learn.
Champ scientifique
Mots‑clés
Programme(s)
Régime de financement
MSCA-ITN - Marie Skłodowska-Curie Innovative Training Networks (ITN)Coordinateur
9712CP Groningen
Pays-Bas