Programme scope and fields

Artificial intelligence, applied mathematics, cognitive sciences, supercomputing, software development   

Acquiring knowledge Knowledge elicitation Data and knowledge retrieval Supervised learning	Simplification and coupling of models Model simplification Model transposition Combining formalisms
Building and transfer of knowledge and models Knowledge building (scientific and know-how) Knowledge transfer Model transfer	Reconstruction of multi-scale dynamics Graphic reasoning Probabilistic reasoning Qualitative reasoning
Decision support Mono- or multi-objective optimisation Trajectory optimisation Automatic or semi-automatic reasoning	Consideration of uncertainties Quantification of uncertainties Representation of uncertainties Propagation of uncertainties

Research fields

•    Knowledge acquisition

All of the methods that make it possible to acquire knowledge, either automatically or not (retrieval of knowledge from data, surveys, interviews, etc.), and to use this knowledge to retrieve relevant information.
 
•    Formalisation and representation of knowledge

A set of tools and processes intended to represent and  organise acquired knowledge with the aim to use and share it. There are many types of formalisms and representations, ranging from the purely symbolic model (qualitative functions, ontology, conceptual graphs, semantic graphs, conceptual maps, mind maps, semantic Web graphs), to the purely qualitative model (e.g., differential equation systems).

•   Knowledge integration

A process consisting of synthesizing knowledge representations at different levels and in different formats (pieces of a puzzle) of a phenomenon into a joint representation.
 
•    Reverse engineering

Development of methods and tools to (1) define the established goal, (2) define arguments that support this goal, and (3) propose itineraries and means to be used to reach this goal.  Reverse engineering therefore includes decision support (multi-criteria, multi-objectives, multi-arguments, multi-stakeholders) and implements methods of control for processes, reverse problems, viability, robustness, etc.       

A parallel approach: application problems and methods

•    Method development

Concerning knowledge acquisition: standardising expert rules and data storage in a software repository

Concerning knowledge representation and reasoning: representing heterogeneous knowledge and taking uncertainties into consideration   

•     Application problems (May 2012)

Eco-design of foods or multi-objective optimisation processes with the Pareto front and decision support:  requiring the development of objective functions representative of technical, usage and environmental properties to be optimised
 
Rational deconstruction of a biomass based on its process planning characteristics: making it necessary to determine the targeted usage of synthons in order to plan the deconstruction process (dry grinding, enzymatic, hydrothermic)

Modelling multistage technical itineraries coupling consecutive stage models, two-by-two, with technical itineraries (multistage process for the multiscale development of a model food, monitoring the quality of a food during its life cycle, etc.)