This program addresses two broad sets of challenges :
The first deals with different facets of the social, economic, legal and ethical acceptability of systems integrating artificial intelligence algorithms.
The second focuses on data challenges. It aims at developing new solutions allowing Il vise à développer des nouvelles solutions permettant :
- the detection and elimination of biases in the selection and interpretation of data used by learning models
- the prevention of the risk of not taking into account rare events in the learning process
- the automatic annotation and semantic representation of heterogeneous, multi-source, multi-scale and time-varying masses of data.
the themes associated with the Acceptable AI program
- AI & society
- Learning with few or complex datas
- Fair Learning
> Discover all themes