MALICE Inria Team

The Inria MALICE team, whose members have a strong expertise in statistical learning, applied mathematics, statistics and optimization, develops algorithmic and theoretical research focused on integrating physical knowledge into machine learning (ML) models.

Leveraging the skills present at the Hubert Curien lab in physics, MALICE aims to foster the development of new methodological contributions in Physics-informed Machine Learning (PiML) with a primary targeted application in Surface Engineering, making possible scientific breakthroughs in both Machine Learning and Physics. Our team focuses on several challenges, including (i) a limited access to training data and the availability of only incomplete background knowledge (typically in the form of Partial Differential Equations - PDEs), (ii) the need of deriving theoretical (generalization, approximation, optimization) guarantees on models learned from both data and physical knowledge and (iii) a strong necessity to transfer knowledge from one dynamical system to another. The advances carried out in machine learning allow to better understand the physics underlying the mechanisms of laser/radiation-matter interaction, enabling to address numerous societal challenges in the fields of space, nuclear, defense, energy or health.

MALICE envisions its activity according to the following three objectives:

1: Theoretical frameworks when learning from data and background knowledge
2: Integration and extraction of knowledge via physics-informed ML models
3: Domain generalization and transfer learning between physical dynamics