About Me

Xinshi Chen (陈心诗)

 
I am a PhD Candidate in Machine Learning at Georgia Tech, advised by Prof. Le Song.

I am broadly interested in principled machine learning. My current research focuses on learning based algorithm design (more theory-oriented), deep learning on structured data (more application-oriented), and their intersections. Besides, I am interested in applications in the area of structural and computational biology. My research is generously supported by Google PhD Fellowship in Machine Learning.

I received my B.S. and M.Phil in Mathematics at the Chinese University of Hong Kong under the supervision of Prof. Eric Chung. I have also spent time at Oak Ridge National Laboratory, Ant Financial, Facebook AI, and MBZUAI as a Research Intern or Research Assistant.

Email: xinshi [dot] chen [at] gatech [dot] edu
Address: CODA Building, 11th Floor, Machine Learning Center, Georgia Tech
My CV

Publication

Preprint & Workshop

A Deep Learning Approach to Recover Conditional Independence graphs
Harsh Shrivastava, Urszula Chajewska, Robin Abraham, Xinshi Chen
NeurIPS 2022 Workshop: New Frontiers in Graph Learning
paper

Graph Condensation via Receptive Field Distribution Matching
Mengyang Liu, Shanchuan Li, Le Song, Xinshi Chen
Arxiv Preprint 2022
paper

Efficient Dynamic Graph Representation Learning at Scale
Xinshi Chen, Yan Zhu, Haowen Xu, Mengyang Liu, Liang Xiong, Muhan Zhang, Le Song
Arxiv Preprint 2021
paper

A Framework For Differentiable Discovery Of Graph Algorithms
Hanjun Dai, Xinshi Chen, Yu Li, Xin Gao, Le Song
NeurIPS 2020 Workshop in Learning Meets Combinatorial Algorithms, Oral
paper | talk

Can Graph Neural Networks Help Logic Reasoning?
Yuyu Zhang*, Xinshi Chen*, Yuan Yang*, Arun Ramamurthy, Bo Li, Yuan Qi, Le Song
NeurIPS 2019 Workshop in Knowledge Representation & Reasoning Meets Machine Learning
paper | poster

Review: Ordinary Differential Equations For Deep Learning
Xinshi Chen
Arxiv Preprent 2019
paper

 

Conference & Journal

Provable Learning-based Algorithm For Sparse Recovery
Xinshi Chen, Haoran Sun, Le Song
International Conference on Learning Representations (ICLR) 2022
paper

Multi-task Learning of Order-Consistent Causal Graphs
Xinshi Chen, Haoran Sun, Caleb Ellington, Eric Xing, Le Song
Advances in Neural Information Processing Systems (NeurIPS) 2021
paper | github | talk | slides

Understanding Deep Architectures With Reasoning Layer
Xinshi Chen, Yufei Zhang, Christoph Reisinger, Le Song
Advances in Neural Information Processing Systems (NeurIPS) 2020
paper | github | talk | slides

Learning To Stop While Learning To Predict
Xinshi Chen, Hanjun Dai, Yu Li, Xin Gao, Le Song
International Conference on Machine Learning (ICML) 2020
paper | github | talk | slides

GLAD: Learning Sparse Graph Recovery
Harsh Shrivastava, Xinshi Chen, Binghong Chen, Guanghui Lan, Srinivas Aluru, Han Liu, Le Song
International Conference on Learning Representations (ICLR) 2020
paper | github | talk

RNA Secondary Structure Prediction By Learning Unrolled Algorithms
Xinshi Chen*, Yu Li*, Ramzan Umarov, Xin Gao, Le Song (*equal contribution)
International Conference on Learning Representations (ICLR) 2020, Oral.
paper | github | talk | slides | news

Efficient Probabilistic Logic Reasoning with Graph Neural Networks
Yuyu Zhang, Xinshi Chen, Yuan Yang, Arun Ramamurthy, Bo Li, Yuan Qi, Le Song
International Conference on Learning Representations (ICLR) 2020
paper | github | talk

Particle Flow Bayes' Rule
Xinshi Chen*, Hanjun Dai*, Le Song (*equal contribution)
International Conference on Machine Learning (ICML) 2019
paper | github | talk | slides | poster

Generative Adversarial User Model for Reinforcement Learning Based Recommendation System
Xinshi Chen, Shuang Li, Hui Li, Shaohua Jiang, Yuan Qi, Le Song
International Conference on Machine Learning (ICML) 2019
paper | github | talk | slides | poster

A distinct class of vesicles derived from the trans-Golgi mediates secretion of xylogalacturonan in the root border cell
Pengfei Wang, Xinshi Chen, Cameron Goldbeck, Eric Chung, Byung-Ho Kang
The Plant Journal 2017
paper

Parametric Finite Element Method for Shape Optimization applied to Golgi Stack
CUHK Theses & Dissertations Collection 2017 [arxiv]
Committee: Prof. Raymond Honfu Chan, Prof. Ronald Lok Ming Lui, Prof. Eric Chung
 

 

Recent Activities

Dec 2021 — in NeuRIPs 2021, I presented our work “Multi-task Learning of Order-Consistent Causal Graphs” virtually. Please check the details here.
Oct 2021 — I passed the PhD proposal defense at GaTech! (On the topic: Duality between Deep Learning and Algorithm Design) Now I am admitted to doctoral candidancy.
Dec 2020 — in NeuRIPs 2020, I presented our work “Understanding Deep Architectures With Reasoning Layer” virtually. Please check the details here.
Nov 2020 — in the HotCSE seminar at GaTech, I presented our work “A Framework For Differentiable Discovery Of Graph Algorithms”. [Introduction]
July 2020 — in 2020 Google PhD Fellowship Summit, I presented our work “Understanding Deep Architectures With Reasoning Layer” virtually. Details here.
June 2020 — in ICML 2020, I presented our work “Learning To Stop While Learning To Predict” virtually. Please check the details here, and stay safe!
April 2020 — in ICLR 2020, three papers are presented virtually. Please check the details here, and stay safe!
June 2019 — in ICML 2019, I presented our works “Particle Flow Bayes’ Rule” [video] and “Generative Adversarial User Model for Reinforcement Learning Based Recommendation System” [video].
May 2019 — I passed the ML PhD qualifying exam at GaTech! (Submitted review on the topic: Ordinary Differential Equations For Deep Learning, and the slides for oral exam.)
March 2019 — in the HotCSE seminar at GaTech, I presented our work “Meta Particle Flow for Sequential Bayesian Inference” [introduction].
Jan. 2019 — in the Scientific Machine Learning Workshop at ICERM, I presented a poster for our work “Particle Flow Bayes’ Rule”.
Nov. 2018 — in the GT MAP Seminar at GaTech, I presented “Sequential Monte Carlo Problem With Mass Transportation”, which is the preliminary version for our paper “Particle Flow Bayes’ Rule”.
July 2018 — in the Workshop on Differential Equations on Networks and Related Problems at Zhejiang University, I presented our work “A no-regret user model”, which is the preliminary version of our paper “Generative Adversarial User Model for Reinforcement Learning Based Recommendation System”.
March 2017 — in the 2017 Imaging Science Camp at USUTech, I presented a part of my thesis “Parametric FEM for shape optimization applied to Golgi stacks”.
Dec 2016 — in the Ceremony of Paul Erdös Award (from World Federation of National Mathematics Competitions) presented to Prof. Kar-Ping Shum [media], I was fortunate to serve as the MC!
Nov 2016 — in the 3rd AoE Symposium on Organelle Biogenesis and Function, I presented our work “A Mathematical Simulation To The Shape Evolution Of Med-/Trans- Golgi Cisternae”, a part of our later published paper.

Biography

M.Phil. (Master of Philosophy) in Mathematics September 2015 - July 2017
• The Chinese University of Hong Kong (Advisor: Prof. Eric Chung)
• Thesis: Parametric Finite Element Method for Shape Optimization [PDF]

B.Sc. in Mathematics September 2011 - July 2015
• The Chinese University of Hong Kong
• Exchange in math department at ETH Zurich, Switzerland January 2014 - June 2014

Experience

Research Assistant Feb 2021 - July 2021
• Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), United Arab Emirates

Research Intern June 2020 - August 2020
• Facebook AI, Menlo Park.

Research Assistant or Teaching Assistant September 2017 - present
• Teach in School of Mathematics, Georgia Institute of Technology.
• Work in Machine Learning Group, supervised by Prof. Le Song.

Research Intern June 2018 - August 2018
• AI department, Ant Financial (affiliate company of Alibaba), Hangzhou, China

Teaching Assistant August 2015 - June 2017
• Department of Mathematics, The Chinese University of Hong Kong

REU Research Intern June 2014 - August 2014
• Oak Ridge National Laboratory, United States
• Mentor: Dr. Joshua Fu, Dr. John Drake and Dr. Kwai Wong
• Solve diffusion-convection equation based on finite element method.
• Title: Modeling Chemical Transport with Galerkin Methods[Project link]

Academic Service

• Program Committee / Reviewer: AAAI 2020-22, ICLR 2020-22, AISTAT 2020-22, MSML 2020-21, ICML 2020-21, NIPS 2020-21, IJCAL 2021

Teaching

School of Computational Science and Engineering, GaTech
• CSE6740 Computational Data Analysis, (Two Guest Lectures), Fall 2019

School of Mathematics, GaTech
• MATH2551 Multivariable Calculus, (Recitation Teaching), Spring 2018 & Fall 2017

Department of Mathematics, CUHK
• MATH3240 Numerical Methods for Differential Equations, (Tutorial), Spring 2016
• MATH3230 Numerical Analysis, (Tutorial), Fall 2016 & Fall 2015
• MATH2010 Advanced Calculus I, (Tutorial), Spring 2016
• MATH1510 Calculus for Engineers, (Tutorial), Fall 2015

Enrichment Programme for Young Mathematics Talents, Hong Kong
• SAYT1054 Mathematical Analysis, (Discussion Group), Fall 2013

Award

• Google PhD Fellowship, 2020-2022
• ICLR Travel Award, 2020
• ICML Travel Award, 2019
• Postgraduate Studentship, CUHK, 2015-2017
• Best oral presentation in 3rd AoE(Area of Excellence) Symposium, 2016
• Professor Charles K. Kao Research Scholarship, 2013-14
• College Head’s list - for outstanding academic performance, 2013-14
• Undergraduate Exchange Scholarship, 2013

Extra-Curriculum

Volunteer Experience
• Bronze Award for Volunteer Service (Individual), 2012, issued by HK Social Welfare Department
• Gold Award for Volunteer Service (Group), 2012, issued by HK Social Welfare Department
• Overall Best Mainland Service Project 2011-12, Caring Heart Community Service Project Scheme

Certificates
• Completion of the Mental Health First Aid Course, certified by MHFA International
• Advanced Open Water Diver, certified by PADI