Buyun Liang

Buyun Liang

Computer Science M.S. Student

University of Minnesota, Twin Cities

Biography

I am a M.S. student in the Department of Computer Science & Engineering at the University of Minnesota, Twin Cities, also work as a graduate research assistant in Prof. Ju Sun’s group.

Broadly, my research is about optimization for machine and deep learning, robustness in vision recognition and AI for science and engineering. Specifically, I am interested in: 1). User-friendly and scalable software for constrained optimization in deep learning; 2). Adversarial robustness in computer vision; 3). AI for topology optimization; 4). Learning from unbalanced biomedical data.

Prior to this, I obtained my first M.S. degree in Materials Science at the University of Minnesota, Twin Cities, and my B.S. degree from the School of Physics, Nanjing University. Before joining Prof. Ju Sun’s group, I worked as a research assistant in Prof. J. Ilja Siepmann’s computational chemistry group.

Download my Curriculum Vitae .

Interests
  • Optimization for Deep Learning
  • Robustness in Vision Recognition
  • AI for Science & Engineering
Education
  • M.S. in Computer Science, 2020 - Present

    University of Minnesota, Minneapolis, MN, USA πŸ‡ΊπŸ‡Έ

  • M.S. in Materials Science, 2018 - 2020

    University of Minnesota, Minneapolis, MN, USA πŸ‡ΊπŸ‡Έ

  • B.S. in Physics, 2014 - 2018

    Nanjing University, Nanjing, Jiangsu, China πŸ‡¨πŸ‡³

Recent News

All news »

[Nov 10, 2022] Our tutorial proposal Deep Learning with Nontrivial Constraints has been accepted to SIAM International Conference on Data Mining (SDM23) Tutorial Program as a two-hour tutorial!

[Oct 20, 2022] Our paper NCVX: A General-Purpose Optimization Solver for Constrained Machine and Deep Learning has been accepted by Neural Information Processing Systems (NeurIPS) Workshop on Optimization for Machine Learning (OPT 2022), see our poster for more details!

[Oct 20, 2022] Our paper Optimization for Robustness Evaluation beyond β„“p Metrics has been accepted by Neural Information Processing Systems (NeurIPS) Workshop on Optimization for Machine Learning (OPT 2022), see our poster for more details!

Publications

(2022). Optimizers Matter in Adversarial Robustness. Under review at IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).

Paper Slides

(2022). NCVX: A General-Purpose Optimization Solver for Constrained Machine and Deep Learning. In Neural Information Processing Systems (NeurIPS) Workshop on Optimization for Machine Learning (OPT 2022).

Paper Poster

(2022). Optimization for Robustness Evaluation beyond β„“p Metrics. In Neural Information Processing Systems (NeurIPS) Workshop on Optimization for Machine Learning (OPT 2022).

Paper Poster

Software

NCVX PyGRANSO
NCVX (NonConVeX) is a user-friendly and scalable python software package targeting general nonsmooth NCVX problems with nonsmooth constraints.
NCVX PyGRANSO