Confiance.ai – Special Session

Confiance.ai (www.confiance.ai) is the largest initiative in Europe for developing a software platform for trustworthy AI engineering. It gathers 9 major industrial companies, two technology research institutes (SystemX and Saint Exupéry) and two top-level scientific organisations (Inria and CEA) within a €45M R&T programme for four years. Moreover, thanks to two subsequent requests for expressions of interest, the programme benefits from the contribution of a dozen of academic laboratories (through PhDs and postdocs) and a similar number of deep tech startups bringing additional expertise.
The challenges addressed in confiance.ai are all scientific and technology challenges contributing to establishing trustworthy AI: (in short) data quality and relevance, data augmentation, robustness, verification and validation, quality assurance, explainability, certification, maintenance in operating conditions, human-machine interaction, etc., all of this with a specific focus on critical systems (life-critical, mission-critical, business-critical) and embedded real-time systems. Key application areas are transportation, defense and security, energy and industrial production, aeronautics and space.

Program

Special Session scheduled on Tuesday Mar 1, 2022 from 10:15 to 11:45 UTC

Time (UTC)	Description
10:15-10:25	The wall of safety for AI: approaches in the confiance.ai program, Bertrand Braunschweig (IRT SystemX), François Terrier (CEA List), Rodolphe Gélin (Renault Group TCR)
10:25-10:40	Can we measure trust?, Agnès Delaborde (LNE, Laboratoire National de Métrologie et d’Essais)
10:40-10:55	Justifying trust in AI/ML system using Engineering Models and Assurance Cases, Eric Jenn (IRT Saint-Exupéry), Morayo Adedjouma (CEA List)
10:55-11:10	How to trust your data: challenges to increase confidence in the data lifecycle of critical systems, Flora Dellinger (Valeo), Camille Dupont (CEA), Xavier Perrotton (Valeo)
11:10-11:15	Questions/Answers
11:15-11:25	Building labelled datasets for real-world tasks with active learning, Thomas Dalgaty (CEA), Fritz Poka Toukam (CEA), Oriane Simeoni (Valeo), Spyros Gidaris (Valeo), Hedi Ben-Younes (Valeo), Nicolas Granger (CEA), Camille Dupont (CEA)
11:25-11:35	An information geometry approach to Randomised smoothing, Hatem Hajri (IRT SystemX), Pol Labarbarie (IRT SystemX)
11:35-11:45	Uncertainty Quantification for Customers Demand Forecasting, Marc Nabhan (Air Liquide)

Special Session Chairs

	Bertrand Braunschweig (IRT SystemX) After a career as a researcher in simulation and artificial intelligence in the oil industry and at the IFP, Bertrand Braunschweig directed the STIC department of the ANR and then the Inria research centres of Rennes and Saclay, and coordinated the national AI research programme. He is now the scientific coordinator of the confiance.ai programme operated by the SystemX technological research institute.
	François Terrier (CEA List) François Terrier is director of the Artificial Intelligence program at CEA. After a PhD in AI and 10 years of R&D in this field, he becomes Director of Research at CEA, he took the lead of the Systems and Software Engineering Department to develop its research activities in model-based engineering and formal analysis of software. Currently, he coordinates the research activities on the new critical challenges for the Responsible Development of Artificial Intelligence: Trust, Frugality and Embedded systems.
	Rodolphe Gélin (Renault Group TCR) Rodolphe Gelin (1965) started his career at CEA (French Atomic Energy Commission), where he worked on robot control for industrial applications and on rehabilitation robotics. He oversaw different teams working on robotics, virtual reality and cognitics. In 2009, he joined SoftBank Robotics to lead the ROMEO project to develop a human size humanoid robot to assist elderly people. Since 2019, he is expert in deep learning at Renault where he drives the activity on trustworthy AI. He is Chairman of the Steering Board of Confiance.ai for FY2022.