Implicit Regularisation, Large Stepsizes and Edge of Stability for (S)GD over Diagonal Linear Networks

Mathieu Even, Scott Pesme, Suriya Gunasekar, Nicolas Flammarion
February 2023
PDF

Abstract

See on arxiv

Type
Conference paper
Publication
Neural Information Processing Systems (NeurIPS), 2023
Mathieu Even
Mathieu Even
PhD student in Statistics and Optimization