“Theory without practice is empty, but equally, practice without theory is blind." ---- I. Kant

Hello! My name is Jingjing Zheng (she/her). My current research interests include efficient training/inference of large models grounded in theory, low-rank/sparse representation learning with applications to AI efficiency, safety & reliability of LLMs under resource constraints. My academic background spans art and design (B.A.), mathematics (M.S. and current Ph.D.), and computer science (completed Ph.D. degree).

Since 2023, I have been pursuing my doctoral studies in Mathematics at the University of British Columbia , under the supervision of Prof. Yankai Cao . In Summer 2024, I undertook a visiting research internship at the Zero Lab, Peking University , where I worked with Prof. Zhouchen Lin on topics related to low-rank-based efficient fine-tuning of large models. In Summer 2025, I co-founded GradientX Technologies Inc., a startup focused on building the next generation of personalized financial intelligence.

Recent News

• [2025] Joined the Organizing Committee of Women and Gender-diverse Mathematicians at UBC (WGM).
• [2025–2026] Appointed to the UBC Green College Academic Committee.
• [2025] Our startup GradientX was selected for the Lab2Market Validate Program(Funded).
• [2025] Two papers accepted to NeurIPS 2025 (see you in San Diego!).

Latest Projects

AdaMSS: Adaptive Multi-Subspace Approach for Parameter-Efficient Fine-Tuning

Jingjing Zheng, Wanglong Lu, Yiming Dong, Chaojie Ji, Yankai Cao^*,^†, Zhouchen Lin^*
(* corresponding authors, † supervisors)
NeurIPS 2025
pdf / code

Differentiable Decision Tree via "ReLU+Argmin" Reformulation

Qiangqiang Mao, Jiayang Ren, Yixiu Wang, Chenxuanyin Zou, Jingjing Zheng, Yankai Cao^*,^†
(* corresponding authors, † supervisors)
NeurIPS 2025 (spotlight)
pdf / code

Handling The Non-Smooth Challenge in Tensor SVD: A Multi-Objective Tensor Recovery Framework

Jingjing Zheng, Wanglong Lu, Wenzhe Wang, Yankai Cao^*,^†, Xiaoqin Zhang^†, Xianta Jiang^†
(* corresponding authors, † supervisors)
ECCV 2024
pdf / code

Bayesian-Driven Learning of A New Weighted Tensor Norm for Tensor Recovery

Jingjing Zheng, Yankai Cao^*,^†
(* corresponding authors, † supervisors)
ICLR 2024@Tiny Papers Track
pdf / code

Structured Sparsity Optimization with Non-Convex Surrogates of L2,0-Norm: A Unified Algorithmic Framework

Xiaoqin Zhang^*,^†, Jingjing Zheng, Di Wang, Guiying Tang, Zhengyuan Zhou, and Zhouchen Lin
(* corresponding authors, † supervisors)
TPAMI 2023
pdf / code