About Me

I am currently a postdoctoral fellow under the guidance of Professor Yong Chen at the University of Pennsylvania. I received my PhD in Computer Science from Emory in 2022, where I was co-supervised by Professors Joyce Ho and Yuanzhe Xi. A main thrust of my research focuses on develop efficient and trustworthy machine learning models for complex real-world data (e.g., tensors, images, texts, graphs, RNA-Seq), under different contexts (e.g., distribution shift, missing data), from various aspects (e.g., accuracy, efficiency, robustness, fairness). Thus far, I have focused on three complementary directions: (1) Scientific Machine Learning (SciML) (e.g., optimization, numerical analysis, parallel computing) (2) Transfer Learning (e.g., Domain adaptation/generalization, Zero-Shot Learning, Parameter efficient fine-tuning) (3) Generative models (e.g., LLMs, GANs, Flow, and Diffusion models)

I am on the job market. Please reach out to me if you find my expertise and research align with your department.

Current research includes but is not limited to the following topics:

• Scientific Mathematics for Machine Learning (e.g., parallel computing, optimization, numerical analysis)
• Transfer learning (e.g., domain adaptation/generalization)
• Generative model for health data

News

• [June-2023] Our work on explainable ML for time series TimeX is accepted by Neurips Spotlight!
• [June-2023] Check out our new preprint on novel efficient nonlinear acceleration nlTGCR!
• [Apr-2023] Our work on time series domain adaptation “Raincoat: Time Series Domain Adaptation under feature and label shifts” is accepted by ICML’23!
• [Feb-2023] New preprint on efficient and robust generative models for EHRs. MedDiff
• [Dec-2022] Thrilled to serve as the CHIL 2023 technology chair! Submit your papers by 2-15-2023.
• [Oct-2022] We are orgnizing a mini-symposium, ‘Optimization for Healthcare’, on SIAM Conference in Optimization 2023
• [Oct-2022] We are orgnizing a mini-symposium, ‘Acceleration methods for scientific and machine learning applications’, on SIAM Conference on Computational Science and Engineering 2023.
• [Oct-2022] New preprint! We are excited to introduce an Efficient Nonlinear Acceleration method that Exploits Symmetry of the Hessian, addressing challenges and providing solutions for nonlinear acceleration.
• [May-2022] “AUTM Flow: Atomic Unrestricted Time Machine for Monotonic Normalizing Flows” is accepted by UAI’22!
• [Apr-2022] Defended my Ph.D. dissertation titled “Acceleration Algorithms for Machine Learning Models”!
• [Jan-2022] “GDA-AM: Solving Minimax via Anderson Mixing” is accepted by ICLR’22!