We are pleased to annouce that one of the papers Zheng Shi co-authored was accepted for presentation at International Conference on Learning Representations (ICLR) 2022. It is a joint work with Majid Jahani, Sergey Rusakov, Peter Richtárik, Michael W. Mahoney, and Martin Takáč.

 
In this paper, they propose a novel algorithm - "OASIS". 
- to solve machine learning problem as well as Deep Neural Networks 
- with guaranteed convergence and strong empirical performance!
- "OASIS" is adaptive with minimum tuning!

Check out the paper here!!!