Predictive models of the SINAPSE project ; optimization by artificial neural networks (ANN)
Résumé
The scientific literature and databases are reliable sources of information concerning the relationships between entities, such as the identification of microorganisms and the metabolites they are capable of producing. However, considering the size of the catalogs of molecules and microorganisms, the available data are very insufficient. Their production remains long and expensive. This is why we propose to design predictive models (neural networks) capable of extrapolating the databases and estimating a probability for the existence of a relationship between two entities. The development must be done with a concern of efficiency, for an operational use.
Financed by CAP 20-25 program from Clermont Auvergne University.