Symposium / Seminar | Informatique, Research

6th SLEIGHT Science Event: Machine Learning

From July 5, 2021 to July 9, 2021

Machine Learning has received a tremendous success during the past years in a large spectrum of scientific fields, far beyond the computer science community, thanks to its ability to address real world problems by learning from data.

The objective of the 6th SLEIGHT Science Event, which took place from the 5th to the 9th of July, 2021, was to allow non-experts in Machine Learning, including master students, PhD students, postdocs as well as permanent researchers, to benefit for their own research from some of the recent advances in Machine Learning.

During this week, the participants had the opportunity to attend 20 hours of lectures, introduced by Florence Garrelie, head of the Graduate School.

After a general introduction by Marc Sebban to the main concepts of statistical machine learning, Charlotte Laclau explained how to anticipate common problems one can face when dealing with real-world data and how to get the data ready for machine learning. Then, Rémi Emonet talked about imbalance learning and anomaly detection, both characterized by a very scarce presence of data of interest in the training samples. Ievgen Redko gave a lecture about domain adaptation and transfer learning. The underlying idea is to benefit from a source task to automatically adapt the learned source model to a so called-target task, different but related to the former. Then, Marc Sebban presented a tutorial on Optimal Transport, a theory that provides a natural geometry for comparing probability measures and which has received a tremendous interest from the machine learning community during the past decade. Finally, Christian Wolf discussed the main models and algorithms for deep learning before presenting Pytorch, one of the most widely used deep learning frameworks.

The attendees benefited also from a joint day between the Manutech-SLEIGHT Graduate School and the doctoral schools MEGA and MATERIAUX (Université de Lyon) with Anthony Gravouil and Francisco Chinesta, who talked about two general topics: Artificial Intelligence and materials. Four companies of the Saint-Etienne Lyon ecosystem shared their experience around the Machine Learning theme.