Ji LI's CV

Ji LI

Computational Mathematical Researcher

Currently

I’m currently a research fellow at Department of Mathematics in National University of Singapore. Before that I was a postdoc at Applied and Computational Mathematics Division in Beijing Computational Science Research Center. Here I develop more efficient algorithms for imaging problem and machine learning applications.

Interest

Applications of supervised and/or unsupervised learning deep learning in inverse problems. Intersection of numerical optimization and practical applications with specific structure. Nonconvex and convex optimization for inverse problems arising in machine learning, computer vision and imaging problem from engineer. Focus on building efficient and scalable algorithms based on the structure and connection between the original nonconvex problem and its convex relaxation formulation.

Education

2019- National University of Singapore Research Fellow (supervisor: Prof. Hui Ji)

2017-19 Beijing Computational Science Research Center Postdoc (supervisor: Prof. Hongkai Zhao).

2012-17 Peking University Ph.D (supervisor: Prof. Tie Zhou). Thesis is entitled Algorithms for Phase Retrieval.

2008-12 Sichuan University Undergraduate student (First class honours)

Grants

2018 National Natural Science Foundation of China (grant No. 11801025)

2017 National Science Foundation for Post-doctoral Scientists of China (grant No. 2017M620589)

Honors

2021 Helsinki Deblur Challenge 2021: the 2nd place

2017 Doctoral thesis: Study of numerical algorithms for phase retrieval

2016 Cushman & Wakefield Scholarship

2016 Special Research Scholarship

2011 National Scholarship for Undergraduate

2010 National Encouragement Scholarship for Undergraduate

Teaching Experiences

2016 Assistant for the Graduate Student Lab, with Linux skills (Peking University, SMS)

2016 Teaching Assistant, Introduction to Programming, Python-based (Peking University, SMS)

2013-15 Teaching Assistant, Mathematical Analysis I, II, III (Peking University, EECS)

Presentations

Tutorials

2017 How to write paper using bibtex and biblatex (PKU, Beijing, China)

Talks

2021 Deep learning approach with conditional blur level information (Inverse Days 2021) 14-16 December, Tampere (online)

2019 Scalable incremental nonconvex optimization approach for phase retrieval (SIAM Conference on Computational Science and Engineering)　24-28 February, Spokane

2018 Scalable incremental nonconvex optimization approach for phase retrieval (16th Conference of CSIAM)　13-16 September, Chengdu

Scalable incremental nonconvex optimization approach for phase retrieval (3nd Jing-Jin-Ji Computational Mathematical Forum) 24-28 August, Hengshui

Scalable incremental nonconvex optimization approach for phase retrieval (Euro 2018) 8-11 July, Valencia

2017 Numerical algorithmic reviews for phase retrieval (Workshop of Mathematical Imaging Computation and Application 2017) 16-19 November, Kaifeng

2016 On the projection method for phase retrieval (International Conference on Information and Computational Science 2016) 2-6 August, Dalian

On the global convergence of generalized phase retrieval (14th Conference of CSIAM) 12-14 August, Xiangtan

2015 Numerical algorithms for generalized phase retrieval (13th IEEE International Conference on Signal Processing) 6-10 November, Chengdu

Preprints

2022 Ji Li, Weixi Wang, Yuesong Nan and Hui Ji. Deep Variational EM Algorithm for Non-uniform Blind Deblurring of Static Scene. Submitted to ECCV2022.

Ji Li, Weixi Wang, Yuesong Nan and Hui Ji. Monte Carlo EM for Self-supervised Blind Image Deconvolution via Langevin Dynamics. Submitted to ECCV2022.

Publications

2022 Weixi Wang, Ji Li and Hui Ji. $\ell_1$-norm Regularization for Short-and-sparse Blind Deconvolution: Point Source Separability and Region Selection. SIAM Journal on Imaging Sciences.

Weixi Wang, Ji Li and Hui Ji. Self-supervised Deep Image Restoration via Adaptive Stochastic Gradient Langevin Dynamics. CVPR2022.

Ji Li, Yuesong Nan and Hui Ji. Un-supervised Learning for Blind Image Deconvolution via Monte-Carlo Sampling. Inverse Problems.

2021 Ji Li. Effective Phase Retrieval of Sparse Signals with Convergence Guarantee. Signal Processing.

2020 Ji Li, and Hongkai Zhao. Solving Phase Retrieval via Graph Projection Splitting. Jul. 2018. accepted by Inverse Problems.

2018 Ji Li, Jian-Feng Cai and Hongkai Zhao. Scalable Incremental Nonconvex Optimization Approach for Phase Retrieval from Minimal Measurements. Jul. 2018. accepted by Journal of Scientific Computing.

Ji Li, Jian-Feng Cai and Hongkai Zhao. Robust Inexact Alternating Optimization for Matrix Completion with Outliers. Sep. 2018. accepted by Journal of Computational Mathematics.

2017 Ji Li, Tie Zhou and Chao Wang. On Global Convergence of Gradient Descent Algorithms for Generalized Phase Retrieval Problem. Aug. 2017. accepted by Journal of Computational and Applied Mathematics.

2016 Ji Li and Tie Zhou. Numerical Optimization Algorithm of Wavefront Phase Retrieval from Multiple Measurements. Mar. 2016. accepted by Inverse Problems and Imaging.

Ji Li and Tie Zhou. On Gradient Descent Algorithm for Generalized Phase Retrieval Problem. Jul. 2016. accepted by Proceedings of ICSP’ 16.

Ji Li and Tie Zhou. On Relaxed Averaged Alternating Reflections (RAAR) Algorithm For Phase Retrieval From Structured Illuminations. Dec. 2016. accepted by Inverse Problems.

Academic Service

Reviewer, IEEE Transactions on Signal Processing

Reviewer, Mathematical Reviews

Reviewer, Journal of Optics

Reviewer, Inverse Problems

Technical skills

  • C/C++
  • Python
  • Matlab
  • LaTeX
  • Bash
  • Git
  • UNIX

Areas of expertise

  • Optimization
  • Statistics
  • Machine learning
  • Data visualisation

Projects

github

My personal github account hosts my hobby development projects as well as listing contributions to open source tools.

blog

myblog

References

Available on request.