Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : LISN/INRIA
Durée : 6 mois
Contact : thibault.monsel@universite-paris-saclay.fr
Date limite de publication : 2024-01-31

Contexte :
This internship is part of a larger project dedicated to building a bridge between Machine Learning and Dynamical Systems : inferring models more robust and less data hungry thanks to physics-based constraints, inspecting the behavior of the models, providing some online guarantees, and relating Physics and computational regularities to improve the model understanding and assessment. The connection between Physics and Machine Learning is nowadays considered in both directions and the scientific construction of this domain is underway. The internship will focus on developing new approachs of modelling dynamical systems as a whole. For the first part of the internship, the intern will get up to speed with continuous-depth models like href{https://arxiv.org/abs/1806.07366}{neural ODE} and href{https://arxiv.org/abs/1904.01681}{augmented Neural ODE}. For the second part of the internship, new research ideas will be explored like href{https://arxiv.org/pdf/2306.14545.pdf}{delayed differential equations}. The candidate is expected to be proactive and have a keen sense of critical thinking. The aim of the internship will be to publish the work in a conference/journal.

Sujet :
Modelling partially observed dynamical systems with continuous-depth models

Profil du candidat :
The candidate should have a solid background in statistics, machine learning and/or applied maths;
knowledge in Python language is required with frameworks like Pytorch/ Tensorflow/JAX. Some background in physics is appreciated too since the intern will train models on datasets from numerical simulations of physical systems. Any knowledge and experience in functional programming is a bonus.

Formation et compétences requises :
The candidate should have a solid background in statistics, machine learning and/or applied maths;

knowledge in Python language is required with frameworks like Pytorch/ Tensorflow/JAX. Some background in physics is appreciated too since the intern will train models on datasets from numerical simulations of physical systems. Any knowledge and experience in functional programming is a bonus.

Adresse d’emploi :
Campus Universitaire bâtiment 650, 1 rue Raimond Castaing, 91190 Gif-sur-Yvette

Document attaché : 202312041200_Offre_Stage_LISN_INRIA_M2.pdf