The aim of this project is to assess the ability of state-of-the-art machine learning tools to provide insight into the physical structure of the cascade process in turbulence. We will consider simplified dynamical models of turbulence, such as shell models, which still capture many key features of real turbulence. The question is whether it is possible to derive an effective model from Navier–Stokes turbulent flow. we hope in this way to get some insights on the physics of the process. To achieve this, we will consider regression techniques as well as Boltzmann machines. This internship is a collaboration between LISN and CEA (SPEC) at Paris-Saclay.