1 paper · 1 filter
Bilel Bensaid, Gaël Poëtte, Rodolphe Turpault
Recently, many machine learning optimizers have been analysed considering them as the asymptotic limit of some differential equations when the step size goes to zero. In other word…