Embedded Files
"Je kunt ook zonder hoop actie ondernemen en zonder succes blijven volharden", Willem de Zwijger, prins van Oranje, graaf van Nassau (1533-1584)
This sentence was at first pronounced in French by Charles le Téméraire, who died in Nancy in 1477 devoured by (probably hungry) wolfes because he was definitively too téméraire...But it is much less funny !
Since September 2017, I am a Professor at the University of Lorraine and member of the Simul Research Group@CRAN.
Since March 2026, I am also member of the European Laboratory for Learning and Intelligent Systems (ELLIS).
I maintain strong links with my former colleagues of Grenoble (but still close friends) and I am still working with APTIKAL Team@LIG.
Since September 2025, I am member of CoCNRS (Section 03) and CID 54
Since January 2026, I am member of the GT ECLAIRE led by E. Devijver (LIG & ELLIS Member) and P.H. Wuillemin (LIP6)
Contact
Centre de Recherche en Automatique de Nancy (CRAN)
Campus Faculté des Sciences
Vandoeuvre les Nancy
e-mail : marianne.clausel@univ-lorraine.fr
Others : ORCID & GoogleScholar
News : Thesis Offer Apply before May, 5th
With E. Devijever (LIG & ELLIS Member), we are offering a PhD thesis funded by INESIA, the new French AI Safey Institute
Towards a framework based on causality and experimental design for AI evaluation
This project aims to build an evaluation framework based on causality and experimental design in order to analyze the robustness of models. The goal is to go beyond static benchmarks by examining how models respond to controlled interventions. We will develop a framework based on causal invariances to identify mechanisms that are truly stable in the face of context changes and distribution shifts. The integration of approaches from experimental design will enable the design of targeted protocols to reveal the weaknesses of models with a minimum of interventions.
These methods will be brought together within INESIA-DoE Lab, a platform dedicated to the generation and execution of experimental designs, accompanied by an open-source library combining causal tools and intervention strategies. Finally, a multi-level analysis will help us understand how local disturbances (in data or architecture) affect overall behavior and contribute to the propagation of errors.
Fell free to contact us for more information :
marianne.clausel@univ-lorraine.fr and emilie.devijver@univ-grenoble-alpes.fr
The Mathematical Foudation of AI one day workshops are BACK on March 18th @SCAI [03/18/2026]
We are still funded by DATAIA ! The AI cluster MIAI will be also part of the organization -:)
Since March 2025, our events are proudly part of the Math & AI network of the Paris region
May 2025-August 2025
I was visiting researcher at Isaac Newton Institute (Cambridge University, UK) for the program "Representing, calibrating & leveraging prediction uncertainty from statistics to machine learning" .
I was awarded a Simons fellowship to fund my visit in Cambridge.
It was a fantastic opportunity to meet world leading researchers and I learnt a lot!
Past grants
French P.I. of the International Project EDDA (Germany-Japan-France). 2021-2024
Current grants
Co-PI with E. Devijver (LIG & ELLIS Member) of the targeted project Causali-T-AI of PEPR IA (U. Lorraine, U. Grenoble-Alpes, INRIA Saclay). 2023-2029
Page updated
Google Sites
Report abuse