Pre-training
Learning at scale, with optimization at the center.
PRIMARY FOCUSFIRST-YEAR PHD STUDENT · COLUMBIA IEOR
Yijin Ren.
Learning at scale.
Thinking in gradients.
Exploring pre-training, first-order methods,
and new questions in reinforcement learning.
01 / ABOUTA little
A little
background.
I am a first-year PhD student in the Department of Industrial Engineering and Operations Research (IEOR) at Columbia University.
I graduated from the Pilot Class of the Research Institute for Interdisciplinary Science at Shanghai University of Finance and Economics (SUFE). During my undergraduate studies, I was advised by Qi Deng at the Antai College of Economics and Management, Shanghai Jiao Tong University, and Yinyu Ye, the K. T. Li Professor of Engineering (Emeritus) in Management Science and Engineering at Stanford University.
My research primarily focuses on pre-training and first-order methods. I am also beginning to explore problems in reinforcement learning.
02 / RESEARCH INTERESTS
Questions I think about.
First-order methods
Understanding and designing methods that learn from gradients.
PRIMARY FOCUSReinforcement learning
Exploring new questions in learning and decision-making.
EXPLORING03 / RESEARCH
Selected publication.
Accelerated Distance-adaptive Methods for Hölder Smooth and Convex Optimization
This paper proposes an adaptive, parameter-free accelerated method for local Hölder smooth convex optimization with optimal rates and no need for Lipschitz or target accuracy tuning.
LET'S CONNECT
Ideas, questions,
or just a hello.
Please feel free to contact me with any questions.