Yuhua Zhu

Yuhua Zhu is a tenure-track Assistant Professor in the Department of Statistics and Data Science at the University of California, Los Angeles (UCLA). Her research focuses on the intersection of partial differential equations (PDEs) and machine learning, with particular emphasis on using differential equations to analyze and design efficient reinforcement learning and optimization algorithms.

Education:

• Ph.D. in Mathematics, University of Wisconsin-Madison (2019), advised by Shi Jin
• M.S. in Mathematics, University of Wisconsin-Madison (2015)
• B.S. in Mathematics, Shanghai Jiao Tong University (2014)
• Postdoc at Stanford University (2019–2022), mentored by Lexing Ying

Research Interests:

Her work spans sequential decision-making (e.g., continuous-time reinforcement learning, multi-armed bandits), optimization (gradient-free methods, asynchronous-SGD, convergence analysis), and uncertainty quantification in kinetic equations (sensitivity analysis, high-dimensional uncertainty). She develops PDE-based frameworks for model-free reinforcement learning and consensus-based optimization methods for distributed learning systems.

Recent Work Trends: Her publications emphasize PDE-driven approaches to reinforcement learning, consensus-based optimization in federated learning, and uncertainty quantification in kinetic equations. Recent topics include addressing adversarial attacks in federated learning, high-order discretization in policy evaluation, and stabilization of kinetic equations with reflective boundary conditions.

Awards & Grants: No specific awards mentioned in the text. Current recruiting focuses on graduate students in reinforcement learning and optimization for machine learning problems.

Labs & Collaborations: Collaborations include work with Jose Carrillo, Shi Jin, Lexing Ying, and others on consensus-based methods and kinetic theory applications in machine learning.