A Unified Approach to Controlling Implicit Regularization Using Mirror Descent
Inspired by the remarkable performance of deep neural networks, understanding the generalization performance of overparameterized models and the effect of optimization algorithms on it has become an increasingly popular question. In particular, there has been substantial effort to characterize the s...
| Auteur principal: | |
|---|---|
| Autres auteurs: | |
| Format: | Thèse |
| Publié: |
Massachusetts Institute of Technology
2023
|
| Accès en ligne: | https://hdl.handle.net/1721.1/151464 https://orcid.org/0000-0002-6203-0198 |