COmbination and DEcision

The CoDe (Combination and Decision) group brings together specialists in tools for representing imperfect information (intervals, fuzzy subsets, probabilities, possibilities, belief functions) and specialists in methods for aggregating and merging heterogeneous information. All the group's skills are federated within the framework of decision support in uncertain dynamic contexts.

Having representations adapted to the nature of information and its imperfections is a crucial point for the design of information fusion systems. In this context, we address the following problems:

  • expression and propagation of uncertainties in physical measurements, using possibilistic approaches, and determination of uncertain parameters in regression models, using fuzzy approaches
  • representation of graded information by fuzzy rules allowing linguistic expression of the model and manipulation of symbolic or numerical quantities adapted to the heterogeneous nature of the data
  • of preference expression in a multi-criteria context to define performance indicators (in a manner consistent with measurement theory) which are then used to drive the improvement of system performance.
  • the formalization of the temporal expression of performance through the structural and temporal decomposition of the objective (the degree of objective attainment)
  • combinations of uncertain information, using Dempster-Shafer's evidential theory to explicitly model the absence of knowledge when sources provide no tangible information
  • fusion of expert data (of an uncertain nature) to assess overall system performance
  • dynamic adaptation of software-intensive systems in response to changes in the environment and/or context, according to user needs.

 

Permanent employees

E. Benoit, L. Berrah, J. Boissière, R. Boukezzoula, V. Clivillé, D. Coquin, L. Foulloy, S. Galichet, M.P. Huget, G. Mauris, S. Perrin, F. Pourraz, L. Valet, H. Verjus