Workshop · November 3–4, 2026 · EDF Lab, Palaiseau
Transfer Learning
and TSFM
Context and Objectives
This event, co-organized by the ANR DECATTLON project and EDF R&D, and sponsored by the French network of ENBIS (frENBIS), is part of the initiatives to promote statistics and machine learning methods within companies and industry. Transfer learning marked a turning point by enabling the reuse of knowledge acquired on one task to solve another, thereby optimizing resources and data. With the emergence of foundation models – general and powerful architectures like Transformers – this approach has undergone a new revolution. These models, pre-trained on massive volumes of data, provide universal bases that are easily adaptable to specific applications through advanced transfer learning techniques. They thus redefine the practices of personalization and efficiency in artificial intelligence.
Speakers
Mathilde Mougeot
Professor of Data Science & Holder of the Industrial Data Analytics and Machine Learning Chair
ENS Paris-Saclay / Centre Borelli / ENSIIE
Mathilde is a researcher in applied mathematics and data science. Her work focuses on statistical learning, predictive modelling, model aggregation, and transfer learning, with strong ties to industrial applications.
Visit website ↗
Shifeng Xie
PhD student
Université Paris Cité
Shifeng is a PhD student in Themis Palpanas's group at Université Paris Cité, after studying engineering at Télécom Paris and Institut Polytechnique de Paris. His research focuses on time series foundation models and agentic systems for forecasting, reasoning, and decision-making.
Visit website ↗
Tahar Nabil
AI Scientist
EDF R&D
Tahar holds a PhD from Télécom Paris in signal processing. At EDF R&D, he works on time series foundation models and generative AI for industrial process design.
Visit website ↗
Adrien Petralia
AI Scientist
Mistral AI
Adrien holds a PhD from Université Paris-Cité on deep learning for time series in the energy sector, spanning signal separation, foundation models, and generative modelling for load disaggregation.
Visit website ↗
Etienne Le Naour
AI Scientist
EDF R&D
Etienne holds a PhD from Sorbonne Université, in collaboration with EDF R&D, on neural representation learning for time series. He now builds foundation models for time series with a focus on forecasting and imputation.
Visit website ↗
Raphael Nedellec
Data Science Manager
Decathlon
Raphael's research focuses on forecasting and calibrated quantile regression using semi-parametric additive models, with widely-used contributions such as the qgam and mgcViz R packages. He now leads data science work at Decathlon.
Visit website ↗Roberto Stanzione
Post-Doc student
INRIA (Valda)
Roberto completed his PhD at the University of Salerno, working on relaxed functional dependencies, concept drift detection in machine learning systems, and federated learning. He is now a post-doc in the Valda team at Inria.
Visit website ↗Join us
The poster session is open for contributions. Submit your abstract before the deadline.