Skip to content

Latest commit

 

History

History
30 lines (21 loc) · 1.32 KB

README.rst

File metadata and controls

30 lines (21 loc) · 1.32 KB

Test status Test coverage Docs status

Problem:Methods with preconditioning with weight decay regularization.
Work type:M1P
Author:Статкевич Екатерина Игоревна
Scientific advisor:Безносиков Александр

Abstract

This work considers a regularization for such algorithms as Adam, OASIS for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The main difference from gradient descent is that Adam's and OASIS algorithm use information about previous gradients to update parameters of the model.

Supplementations

  1. Link review.