Lucas Journel (LJLL) Non-convexe optimisation using stochastic processes.
L’exposé aura lieu en salle de séminaire, dans la limite des places disponibles, et sera aussi retransmis au lien BigBlueButton suivant :
Code d’accès : 978985
Voici l’abstract de l’exposé :
“In many fields such as statistics or molecular dynamics, optimizing functions with multiple local minima, not necessarily confined within a compact set, poses a significant challenge. Traditional gradient-descent algorithms tend to converge to local minima, necessitating a broader exploration of the solution space to identify the global minimum. However, exhaustive search methods are impractical in high-dimension. In this talk, I will introduce two stochastic processes designed to address this optimization problem”