Master Internship - Graph Inference in Dynamical Systems

Contexte et atouts du poste

The aim of this internship is to investigate the inference of graphs to describe the behavior of dynamical systems (e.g., time-series from climate database).

Subject: State-space models (SSMs) are common tools in time-series analysis for inference and prediction in dynamical systems. SSMs are versatile probabilistic models that allow for Bayesian inference by describing a (generally Markovian) latent process. However, the parameters of that latent process are often unknown and must be estimated. In [1,2], we have proposed an innovative approach to perform the parameters inference as sparse graphs. The approach, based on an Expectation-Maximization mechanism and advanced non-smooth optimization tools, provides promising results, and benefits from sound convergence guarantees. However, it is limited to the class of linear Gaussian SSMs with first-order Markovian dependancies. In this internship, we plan to explore exten...