Talk
Treatment strategies are critical in healthcare, particularly when outcomes are subject to censoring. This study introduces the Counterfactual Buckley–James Q-Learning framework, which integrates counterfactual reasoning with the Buckley–James method and reinforcement learning to address challenges arising from longitudinal survival data. The Buckley–James method imputes censored survival times via conditional expectations based on observed data, offering a robust mechanism for handling incomplete outcomes. By incorporating these imputed values into a counterfactual Q-learning framework, the proposed method enables the estimation and comparison of potential outcomes under different treatment strategies. This facilitates the identification of optimal dynamic treatment regimes that maximize expected survival time. Through extensive simulation studies, the method demonstrates robust performance across various sample sizes and censoring scenarios, including right censoring and missing at random.Application to real- world clinical trial data further highlights the utility of this approach in informing personalized treatment decisions, providing an interpretable and reliable tool for optimizing survival outcomes in complex clinical settings.

Bio
Dr. Jong-Min Kim (Born in South Korea) is Professor of Statistics at University of Minnesota-Morris, USA. He received his PhD (Statistics) in 2002 from the Department of Statistics, Oklahoma State University, USA (Minor: Mathematics). He worked as Research Fellow at SAMSI -The Statistical and Applied Mathematical Sciences Institute (NSF, Duke, NCSU and UNC). He received the Morris Faculty Distinguished Research Award. He also joined University of Minnesota Data Science Initiative Core Member in May 2024 and joined EGADE Business School Tecnologico de Monterrey as Adjunct Professor of Finance in June 2025. He is in the American Statistical Association—Outstanding Statistical Application Award Committee (2026- 2027), General Director (2023-2024) and Board Member (2023-2028) of Korean International Statistical Society, Council of Chapter Representatives (2021-2025) American Statistical Association-Twin Cities Chapter, and Statistics Discipline Coordinator, University of Minnesota at Morris. He is Elected Member of the International Statistical Institute (ISI), October 2005.

Email
jongmink@morris.umn.edu

Talk
This talk is based on a series of joint works with Zhen Chao, Harris Cobb, Haoxiang Huang, Weiming Ding, Hwi Lee, Tzu Jung Lee, Dexuan Xie, and Vigor Yang. We have developed a neural network approach capable of accurately predicting both shock interactions and smooth regions of solutions to the Euler equations in 1D and 2D. For example, to predict the solution of the 1D Euler equations at a specific space-time location, the output of a neural network can be designed to provide the solution value at that location. However, if the input consists solely of a low-cost numerical solution patch within a local domain of dependence, the neural network lacks the ability to distinguish between inputs spanning a shock and those within a smooth region. Our approach leverages two numerical solutions from a converging sequence—computed using low-cost numerical schemes within the local domain of dependence of a given space-time location. These serve as the input for the neural network to generate high-fidelity solution at the location. Despite the smeared nature of input solutions, the resulting output provides sharp approximations for solutions containing shocks and contact discontinuities. This method dramatically reduces computational complexity compared to fine-grid numerical simulations, achieving at least a two-order-of-magnitude reduction in 2D, with potential for even greater savings in higher dimensions due to its localized methodology. Moreover, the training data requirement remains minimal, as a single fine-grid simulation can generate hundreds or thousands of local samples for training. The method sustains strong generalization even when confronted with complex and pronounced singularities in the solutions. Beyond its efficiency, this approach naturally extends to complex domain geometries, enabling training on one domain geometry while facilitating predictions on another. We will also discuss extensions of this methodology to unstructured grids, electromagnetic wave scattering off curved conductors with corners, and the Poisson–Nernst–Planck ion channel model.

Bio
Yingjie Liu earned his Bachelor's and Master's degrees in Computational Mathematics from Peking University in 1987 and 1990, respectively, and obtained his Ph.D. in Applied Mathematics from the University of Chicago in 1999. He was a postdoctoral researcher in the Department of Applied Mathematics and Statistics at SUNY Stony Brook from 1999 to 2002. Since 2002, he has served as a professor—progressing from assistant to associate to full—in the School of Mathematics at the Georgia Institute of Technology. His research focuses on the development and analysis of numerical methods for solving differential equations, particularly partial differential equations. He has developed or co-developed several innovative numerical methods, some of which are widely used in both industry and academia. These methods include: (1) the Back-and-Forth Error Compensation and Correction (BFECC) method for convection and hyperbolic systems; (2) central schemes and central discontinuous Galerkin methods on overlapping cells; (3) Hierarchical Reconstruction (HR) limiting methods for eliminating spurious numerical oscillations while preserving the order of approximation; and (4) Neural Networks with Local Converging Inputs (NNLCI), a general approach for solving PDEs with orders-of-magnitude complexity reduction and minimal training data requirements.

Email
yingjie@math.gatech.edu

Talk
The functions under approximation here have as a domain a finite dimensional Banach space with dimension N 2 N and are with values in RN. Exploiting some topological properties of the above we are able to perform multi-composite general Neural Network multivariate approximation to the above functions. The treatment is quantitative. We produce multivariate multi-composite general Jackson type inequalities involving the modulus of continuity of the function under approximation. The established convergences are pointwise and uniform. Our technique is expected to lead to accelerated speeds of convergence.

Bio
George A. Anastassiou was born in Athens, Greece in 1952. He received his B.SC degree in Mathematics from Athens University, Greece in 1975. He received his Diploma in Operations Research from Southampton University, UK in 1976. He also received his MA in Mathematics from University of Rochester, USA in 1981. He was awarded his Ph. D in Mathematics from University of Rochester, USA in 1984. During 1984-86 he served as a visiting assistant professor at the University of Rhode Island, USA. Since 1986 till 2024, he had been a faculty member at the University of Memphis, USA. He had been a full Professor of Mathematics since 1994. Since August 2024 is a Professor Emeritus. His research area is “Computational Analysis” in the very broad sense. He has published over 750 research articles in international mathematical journals and over 58-monographs, proceedings and textbooks in well-known publishing houses. Several awards have been awarded to George Anastassiou. In 2007 he received the Honorary Doctoral Degree from University of Oradea, Romania. He is associate editor in over 93 international mathematical journals and had been editor in-chief in 3 journals, most notably in the well-known “Journal of Computational Analysis and Applications”.

Email
ganastss2@gmail.com