Skip to main content

Shixin Xu

Assistant Professor of Mathematics at Duke Kunshan University
DKU Faculty

Overview


Shixin Xu is an Assistant Professor of Mathematics whose research spans several dynamic and interconnected fields. His primary interests include machine learning and data-driven models for disease prediction, multiscale modeling of complex fluids, neurovascular coupling, homogenization theory, and numerical analysis. His current projects reflect a diverse and impactful portfolio:

  • Developing predictive models based on image data to identify hemorrhagic transformation in acute ischemic stroke.
  • Conducting electrodynamics modeling of saltatory conduction along myelinated axons to understand nerve impulse transmission.
  • Engaging in electrochemical modeling to explore the interactions between electric fields and chemical processes.
  • Investigating fluid-structure interactions with mass transport and reactions, crucial for understanding physiological and engineering systems.

These projects demonstrate his commitment to addressing complex problems through interdisciplinary approaches that bridge mathematics with biological and physical sciences.

Current Appointments & Affiliations


Assistant Professor of Mathematics at Duke Kunshan University · 2019 - Present DKU Faculty
Assistant Professor of the Practice of DKU Studies at Duke University · 2024 - Present DKU Studies

Recent Publications


An imbalanced learning-based sampling method for physics-informed neural networks

Journal Article Journal of Computational Physics · August 1, 2025 This paper introduces Residual-based Smote (RSmote), an innovative local adaptive sampling technique tailored to improve the performance of Physics-Informed Neural Networks (PINNs) through imbalanced learning strategies. Traditional residual-based adaptive ... Full text Cite

Improving GBDT performance on imbalanced datasets: An empirical study of class-balanced loss functions

Journal Article Neurocomputing · June 14, 2025 Class imbalance poses a persistent challenge in machine learning, particularly for tabular data classification tasks. While Gradient Boosting Decision Trees (GBDT) models are widely regarded as state-of-the-art for these tasks, their effectiveness diminish ... Full text Cite
View All Publications

Education, Training & Certifications


University of Science and Technology of China (China) · 2013 Ph.D.

External Links


Personal Webpage