Syracuse University
Applied Mathematician
Ph.D. Candidate in Mathematics
Statistics emphasis
01 / About
AI has made this a golden age for mathematics, and the best time in history to be an applied mathematician.
I am an applied mathematician and machine learning researcher pursuing a Ph.D. at Syracuse University. I combine mathematical modeling and numerical optimization with AI-assisted implementation and experimentation to turn research ideas into working methods.
I develop data-efficient learning methods for image restoration and inverse problems. My work ranges from transferring supervision across image denoisers to restoring images without an external training set.
Research track recordFirst and corresponding author of HyperDn, accepted to NeurIPS 2026 (Main Track); co-first and corresponding author of MG-SpaIR, published in the Journal of Mathematical Imaging and Vision.
Open to Research Scientist / Applied Scientist roles and research collaborations in machine learning, computer vision, and computational imaging.
Syracuse University
Ph.D. in Mathematics · Statistics emphasis
2022–2027 (expected)
Sun Yat-sen University
M.S. in Computational Mathematics
2019–2022
Sun Yat-sen University
B.S. in Information and Computing Science
2015–2019
NeurIPS 2026 · Main Track · Accepted
First and corresponding author
HyperDn transfers oracle supervision across denoiser configurations, reducing the target labels needed to predict hyperparameters for new image-denoising tasks.
Journal of Mathematical Imaging and Vision · 2026
Volume 68, Article 48 · Co-first and corresponding author
Image restoration from a single degraded observation, combining multi-grade implicit neural representations with sparse regularization.
arXiv preprint · 2023
arXiv:2308.11904 · First author
A successive two-stage optimization method with support alteration for sparse generalized eigenvalue problems.
MATLAB · Sparse optimization
I implemented and benchmarked a support-alteration solver for sparse PCA, Fisher discriminant analysis, and canonical correlation analysis, on problems with up to 20,000 variables.
Tools I work with
Python · PyTorch · C++ · SQL · MATLAB · NumPy · pandas · scikit-learn · Linux · Git · HTCondor
A series of essays on how to choose, sustain, and evaluate research in the AI era.
Essay 02 · Coming soon
Essay 01 · September 28, 2026
Why maximizing intrinsic research value can require an initial phase devoted entirely to external return.
“You can fool some of the people all of the time, and all of the people some of the time, but you cannot fool all of the people all of the time.”
All I need to understand about the Navier–Stokes problem: the equations are not a particularly good model.
“An approximate answer to the right problem is worth a good deal more than an exact answer to an approximate problem.”
“…general methods that leverage computation are ultimately the most effective, and by a large margin.”
For Research Scientist / Applied Scientist opportunities and research collaborations, please reach me by email.