TY  - RPRT
TI  - A Markov Chain Theory Approach to Characterizing the Minimax Optimality of Stochastic Gradient Descent (for Least Squares)
AU  - Prateek Jain
AU  - Sham M. Kakade
AU  - Rahul Kidambi
AU  - Praneeth Netrapalli
AU  - Venkata Krishna Pillutla
AU  - Aaron Sidford
PY  - 2018
DO  - 10.4230/lipics.fsttcs.2017.2
UR  - https://arxiv.org/abs/1710.09430
ID  - 1710.09430
ER  -