CV

Education

• B.S. in Mathematics and minor in Economics, Peking University, July 2013
• M.A. in Statistics, Columbia University, Dec. 2014
• Ph.D. in Computer Science, Columbia University, 2020 (expected)
• Relevent courseworks: Topics in interactive learning, Natural Language Processing, Advanced Machine Learning, Advanced Algorithm, Analysis of Algorithms I-II, Topics in Advanced Probabilities, Graphical Model, Probability Theory I-II, Statistical Inference Theory I-III, Linear Regression Model, Machine Learning, Nonparametric Statistics.

Honors and Awards:

• Inaugural Cheung-Kong Graduate School of Business (CKGSB) Fellowship, July 2018
• Kathryn and Shelby Cullom Davis International Fellowship, Jan. 2015
• Silver Medalist in National Undergraduate Mathematical Contest, Nov. 2011
• Silver Medalist in Chinese Mathematical Olympiad Winter Camp, Jan. 2009
• Gold Medalist in National Chinese Mathematical Olympiad in Senior, Nov. 2008

Skills

• Programming skill: R, Python

Publications

Global analysis of Expectation Maximization for mixtures of two Gaussians

Ji Xu, Daniel Hsu, Arian Maleki. "Global analysis of Expectation Maximization for mixtures of two Gaussians." Neural Information Processing Systems (NeurIPS), 2016 (Oral Presentation + Travel Award).

Phase retrieval based on spectral initialization and approximate message passing

Ji Xu, Junjie Ma, Arian Maleki. "Phase retrieval based on spectral initialization and approximate message passing." In preparation, 2018.

Approximate message passing for amplitude-based optimization.

Junjie Ma*, Ji Xu*, Arian Maleki. "Approximate message passing for amplitude-based optimization." International Conference on Machine Learning (ICML), 2018 (Long Talk + Travel Award).

Risk estimation of leave-one-out cross validation (LO) in high dimension regime via approximate message passing (AMP).

Ji Xu, Arian Maleki, Kamiar Rahnama Rad, Daniel Hsu. "Risk estimation of leave-one-out cross validation (LO) in high dimension regime via approximate message passing (AMP)." In submission to Annals of Statistics , 2018.

Benefits of over-parameterization with EM

Ji Xu, Daniel Hsu, Arian Maleki. "Benefits of over-parameterization with EM." Neural Information Processing Systems (NeurIPS) , 2018.

Optimization-based AMP for Phase Retrieval: The Impact of Initialization and $\ell_2$-regularization

Junjie Ma, Ji Xu, Arian Maleki. "Optimization-based AMP for Phase Retrieval: The Impact of Initialization and $\ell_2$-regularization" IEEE Transactions on Information Theory 65 (6), 3600-3629 , 2019.

Two models of double descent for weak features.

Mikhail Belkin, Daniel Hsu, Ji Xu. "Two models of double descent for weak features." In submission to IEEE Transactions on Information Theory , 2019.

Spectral Method for Phase Retrieval: an Expectation Propagation Perspective

Junjie Ma, Rishabh Dudeja, Ji Xu, Arian Maleki, Xiaodong Wang. "Spectral Method for Phase Retrieval: an Expectation Propagation Perspective." Under review of IEEE Transactions on Information Theory , 2019.

On the number of variables to use in principal component regression

Ji Xu, Daniel Hsu. "On the number of variables to use in principal component regression" To appear at Neural Information Processing Systems (NeurIPS) , 2019.

• *Equal contribution

Presentations

Global analysis of Expectation Maximization for mixtures of two Gaussians

December 06, 2016

Talk at NIPS 2016 conference, Barcelona, Spain

Global analysis of Expectation Maximization for mixtures of two Gaussians

July 12, 2019

Talk at ICML 2018 conference, Stockholm, Sweden