top of page

HELLO, I'M

​Keyi Chen

IMG_3069.jpg

​Keyi Chen

​Hi, I am currently a PhD student in Boston University, Computer Science Department, advised by Francesco Orabona​. Previously, I was at Stony Brook University.

My research focus on online learning, theoretical machine learning with special interest on parameter-free algorithms that does not require tuning learning rate.

Skills & Languages

Education

Publications

2018 - Now

Boston University – Boston, MA; 

PhD; Computer Science

​

  • Keyi Chen, Francesco Orabona, Generalized implicit follow-the-regularized-leader. ICML 2023

  • Keyi Chen, Francesco Orabona, Implicit Interpretation of Importance Weight Aware Updates, ICML 2023 Workshop

  • Keyi Chen, Ashok Cutkosky, Francesco Orabona, Implicit Parameter-free Online Learning with Truncated Linear Models, ALT 2022

  • Keyi Chen, John Langford, Francesco Orabona, Better Parameter-free Stochastic Optimization with ODE updates for Coin-betting, AAAI 2022

2016–2018

Stony Brook University – Stony Brook, NY

MS; Statistics

Academic Service

2011–2015

Shandong University, Shandong, China 

BS; Statistics

  • Reviewer, AISTATS 2022/2023, AAAI 2023/2024​

Work Experience

Research Assistant, Optimal Lab at Boston University
Advisor: Francesco Orabona

​Jan.2019 - Now

  • Designed algorithms for convex optimization, performed theoretical analysis, with special interest on online learning, and parameter-free algorithms which does not require tuning learning rate

  • Implemented new algorithms and conduct comparison with baseline algorithms in Python/ Matlab to evaluate their empirical performance on classification/regression machine learning tasks 

Machine Learning Intern, IQVIA

​Jun 2022 - Aug 2022

  • Built machine learning models to predict personalized output in medical area tasks, extended the personalized input output hidden markov model (PIOHMM) to Bernoulli output    

  • Implemented a PIOHMM that captures personalized effect, and nurse effect, to predict the probability that the patient will engage in the future at specific period of time

bottom of page