The image semantic interpretation has become a major challenge for research since the massive proliferation of data sources. However, the classifiers used in this field are often black boxes, and it is especially true with the current trend of deep learning. However, in some applications, such as medical or security applications, having a good classification rate is not enough: we must be able to prove and thus explain the decision. It is in this context that we offer these work that involves the study and design of interpretable classification models. In the context of semantic image interpretation, we will focus our work on learning fuzzy spatial rules for use in an expert system. The spatial and contextual information is indeed an essential information for understanding and interpreting the content of an image.

This position is open until it is filled.

Département: DM2I (LIST)
Laboratory: Laboratoire d’Analyse des Données et d’Intelligence des Systèmes
Start Date: 01-10-2015
ECA Code: SL-DRT-15-1047
Contact: jean-philippe.poli<στο>cea.fr