Events

Nov 15 - Nov 9, 2025

• 

  AI and Mathematics: Efficient Machine Learning Algorithms Inspired by Dynamical Systems, Complex Analysis, and Embedding Theory

  Zhihong Xia, Chair Professor, Department of Mathematics, Southern University of Science and Technology

  Nov 11, 10:30 - 11:30

  B9 L3 R3120

  AI machine learning PDEs Dynamical Systems scientific computing

  This talk describes a novel machine learning algorithm, inspired by complex analysis and dynamical systems theory, that significantly improves efficiency in solving partial differential equations and scientific computing, while providing a theoretical framework for reconstructing complex, unknown systems from partial observational data.
• 

  Towards Flexible Polyhedral Nets

  Pirahmad Olimjoni, Ph.D. Student, Applied Mathematics and Computational Science

  Nov 10, 14:30 - 15:30

  B4 L5 R5209

  Isotropic Geometry discrete differential geometry applied mathematics algorithms optimization

  This thesis research provides a comprehensive classification of flexible geometric nets of arbitrary size in both Euclidean and isotropic geometries, revealing that only two distinct classes exist in each setting and demonstrating their relationship to mechanical design.

Nov 8 - Nov 2, 2025

• 

  Proper Random Walk Spline Models

  Eman Kabbas, Ph.D. Student, Applied Mathematics and Computational Science

  Nov 2, 15:00 - 17:00

  B3 L5 R5209

  Bayesian and computational Statistics data science

  This dissertation introduces the Proper Random Walk of order 2 (PRW2), a full-rank Gaussian Markov random field that provides a principled alternative to intrinsic random walk (RW2) priors. By construction, RW2 models exhibit heteroscedastic marginal variances, inflated boundary effects, sensitivity to grid design, and unbounded forecast uncertainty—features that undermine the reliability of inference, particularly in sparse-data settings or beyond the observed domain.

Nov 1 - Oct 26, 2025

• 

  Event Status: Cancelled | Accelerating Branch-and-Bound Graph Algorithms with GPUs

  Izzat El Hajj, Assistant Professor, Computer Science, American University of Beirut

  Oct 30, 12:00 - 13:00

  B9 L2 R2325

  This talk presents multiple techniques that we have developed to load balance the search tree traversal on GPUs and mitigate the strain on memory capacity and bandwidth.
• 

  Unlocking Euclidean Problems with Isotropic Initialization

  Mikhail Skopenkov, Research Scientist, Computer Science

  Oct 30, 12:00 - 13:00

  B9 L2 R2325

  The seminar introduces a novel, general approach for solving challenging constraint systems in Euclidean geometry problems by leveraging analogous, structure-preserving simplifications found in isotropic geometry to initialize and guide optimization algorithms.
• 

  Ringleader ASGD: The First Asynchronous SGD with Optimal Time Complexity under Data Heterogeneity

  Arto Maranjyan, Ph.D. Student, Computer Science

  Oct 28, 14:30 - 15:30

  B1 L3 R3119

  This talk introduces Ringleader ASGD, the first asynchronous SGD algorithm that attains the theoretical lower bounds for parallel first-order stochastic methods in the smooth nonconvex regime, thereby achieving optimal time complexity under data heterogeneity and without restrictive similarity assumptions.
• 

  Collective Behaviors - From Traffic Congestion to Chase-and-Escape

  Toru Ohira, Professor, Dean of the Graduate School of Mathematics, Nagoya University, Japan

  Oct 26, 14:30 - 15:30

  B5 L5 R5220

  Mathematical modeling

  In this lecture, we introduce representative mathematical models of collective motion, showing how they capture unexpected emergent patterns and discuss chase-and-escape dynamics, a classic topic in mathematical modeling.

Oct 25 - Oct 19, 2025

• 

  A Generalized Gaussian Min–Max Framework Inferring Post-Processing Effects in High-Dimensional Estimation

  Abla Kammoun, Senior Research Scientist, Electrical and Computer Engineering

  Oct 21, 14:30 - 15:30

  B1 L3 R3119

  statistical analysis

  This talk introduces the Convex Gaussian Min–Max Theorem (CGMT) as a principled framework to analyze the performance of such solutions, and presents a new Gordon-type inequality to study functionals of the solutions that arise when direct deployment is infeasible.

Oct 18 - Oct 12, 2025

• 

  Bias-Reduced Estimation of Structural Equations Models

  Haziq Jamil, Research Specialist, Statistics

  Oct 16, 12:00 - 13:00

  B9 L2 R2325

  This talk demonstrates that the reduced-bias M-estimation (RBM) framework is a computationally efficient and robust method for mitigating finite-sample bias in structural equation models, outperforming standard estimators, especially in small-sample contexts.

Oct 11 - Oct 5, 2025

• 

  Block Low-Rank Matrices for Modern Scientific Computing

  Daria Sushnikova, Postdoctoral Research Fellow, Computer, Electrical and Mathematical Sciences and Engineering

  Oct 9, 12:00 - 13:00

  B9 L2 R2325

  Block low-rank matrices provide framework for compressing and accelerating large-scale computations. In this talk, I will introduce the basic principles behind these matrix formats, highlight notable algorithms that exploit their structure, and discuss their growing role in modern computational mathematics. Applications ranging from seismic imaging and computational biology to artificial intelligence will be used to illustrate the broad impact of block low-rank methods on science and engineering.

Oct 4 - Sep 28, 2025

• 

  AI for Science: AI-Driven Knowledge Discovery

  Dongxiao Zhang, Chair Professor, Faculty of Engineering; Provost and Executive Vice President, Eastern Institute of Technology, Ningbo (EIT)

  Oct 2, 14:30 - 15:30

  B1 L3 R3119

  AI machine learning

  This talk focuses on the latest progress in this area, often called AI for Science, and specifically on how AI is revolutionizing scientific exploration and the uncovering of fundamental physical laws.
• 

  Event Status: Cancelled | Efficient Bayesian Methods for Biostatistics

  Janet van Niekerk, Research Scientist, Statistics

  Oct 2, 12:00 - 13:00

  B9 L2 R2325

  bayesian methods

  In this talk, I will present some case studies where we approach near real-time inference for complex Biostatistics models, such as disease mapping and brain activation mapping models, among others often encountered in the biostatistics domain, using INLA.
• 

  Towards Finite Element Tensor Calculus

  Kaibo Hu, Associate Professor, Mathematical Institute, University of Oxford

  Sep 30, 14:30 - 15:30

  B1 L3 R3119

  Finite elements tensor calculus PDEs

  In this talk we provide an overview of some efforts to develop finite elements for tensors, motivated by the need for structure-preserving and compatible discretization.
• 

  New Challenges in the Kinetic Theory of Complex Fluids

  Umberto Zerbinati, Stipendiary Lecturer, Oriel College, University of Oxford

  Sep 29, 16:00 - 17:00

  B1 L4 R4214

  PDEs numerical PDEs Kinetic Theory

  This seminar presents a new kinetic theory for complex fluids, detailing the development of a Boltzmann-type equation and discussing its theoretical properties concerning relaxation to equilibrium, hydrodynamic limits, and a novel energy functional for calamitic fluids.

Sep 27 - Sep 21, 2025

• 

  Nonparametric Functional Quantiles, Skewness, and Probability Bands via the Functional Signed Directionality

  Emmanuel Ambriz, Postdoctoral Research Fellow, Statistics

  Sep 25, 12:00 - 13:00

  B9 L2 R2325

  functional signed directionality

  We propose the Functional Signed Directionality (FuSD), a nonparametric order for function-valued random elements that defines a pushforward distribution on the real line. This construction enables probabilistic inference in the functional domain: we define set-valued functional quantiles, introduce quantile-based summaries of functional spread and skewness, and construct tight probability bands that reflect the underlying functional law via the FuSD distribution.

Sep 20 - Sep 14, 2025

• 

  Neural Methods for Amortized Inference with Models for Spatial Extremes

  Raphaël Huser, Associate Professor, Statistics

  Sep 18, 12:00 - 13:00

  B9, L2, R2325

  Neural Bayes estimators are neural networks that approximate Bayes estimators. Once trained, these estimators are not only statistically efficient, but also extremely fast to evaluate and amenable to rapid uncertainty quantification. Neural Bayes estimators thus have compelling advantages when used with spatial models that have a computationally intractable likelihood function, as often the case when modeling spatial extremes. In this talk, I will showcase the power of neural Bayes estimators for spatial extremes in a range of climate-related and geo-environmental data applications.
• 

  From the Ball-Proximal (Broximal) Point Method to Efficient Training of LLMs

  Peter Richtarik, Professor, Computer Science

  Sep 16, 16:00 - 17:00

  B1 L3 R3119

  AI machine learning optimization algorithms LLM

  This talk introduces the Ball-Proximal Point Method, a new foundational algorithm for non-smooth optimization with surprisingly fast convergence, and Gluon, a new theoretical framework that closes the gap between theory and practice for modern LMO-based deep learning optimizers.

Sep 13 - Sep 7, 2025

• 

  Gaussian Random Fields on Metric Graphs

  David Bolin, Professor, Statistics

  Sep 11, 12:00 - 13:00

  B9 B2 L2325

  Gaussian random fields Metric graphs Statistical Modeling

  This talk presents a comprehensive mathematical and statistical theory, along with user-friendly software, for modeling data with Gaussian random fields on metric graphs by developing valid covariance functions based on network distance.

Sep 6 - Aug 31, 2025

• 

  From the Ball-Proximal (Broximal) Point Method to Efficient Training of LLMs

  Peter Richtarik, Professor, Computer Science

  Sep 4, 12:00 - 13:00

  B9 L2 R2325

  AI machine learning optimization algorithms LLM

  This talk introduces the Ball-Proximal Point Method, a new foundational algorithm for non-smooth optimization with surprisingly fast convergence, and Gluon, a new theoretical framework that closes the gap between theory and practice for modern LMO-based deep learning optimizers.

Jul 19 - Jul 13, 2025

• 

  Randomized Greedy Algorithms for Neural Network Optimization in Solving Partial Differential Equations

  Xiaofeng Xu, Ph.D. Student, Applied Mathematics and Computational Science

  Jul 15, 17:00 - 19:00

  B4 L5 R5220

  PDEs optimization machine learning randomized orthogonal greedy algorithm

  This thesis introduces the randomized orthogonal greedy algorithm (ROGA) to bridge the gap between theoretical and practical performance of shallow neural networks for solving partial differential equations by overcoming key optimization challenges to achieve provably optimal convergence rates.

Jul 12 - Jul 6, 2025

• 

  Discovery of Low-Dimensional Generative Models for Complex Dynamical Systems

  Juan Pablo Muñoz Díaz, Ph.D. Student, Applied Mathematics and Computational Science

  Jul 9, 09:00 - 10:00

  B2 L5 R5220; Zoom Meeting 95274807609

  generative ai applied mathematics bioscience Dynamical Systems deep learning

  This thesis presents a data-driven framework for discovering low-dimensional generative models of complex systems by using a library of normal-form equations to identify both observable dynamics and hidden control variables directly from time-series data.
• 

  Theory and Implementation of Novel Numerical Methods for Multiphysics Interface Problems

  Najwa Alshehri, Ph.D. Student, Applied Mathematics and Computational Science

  Jul 8, 15:00 - 17:00

  B2/B3 L0 R0215

  finite element method numerical analysis fluid-structure interactions mixed finite elements

  This thesis develops and analyzes Finite Element Methods (FEM) for multiphysics interface problems using the Fictitious Domain with Distributed Lagrange Multiplier (FD-DLM) framework. It introduces new families of stable mixed methods with discontinuous Lagrange multiplier spaces, studies both a priori and a posteriori error estimates, and designs multigrid preconditioners. Theoretical results are supported by numerical experiments.

May 24 - May 18, 2025

• 

  Stochastic Numerics and Statistical Learning: Theory and Applications Workshop 2025

  May 18, 09:00 - May 25, 15:30

  Auditorium 0215 between B2 & B3

  stochastic algorithm statistical learning machine learning

  Explore the latest in stochastic algorithms, statistical learning and optimization at KAUST’s Stochastic Numerics and Statistical Learning Workshop 2025.

May 17 - May 11, 2025

• 

  Perturbation Methods in PDE: from Heuristics to Breakthroughs in Regularity Theory - Part 3

  Dr. Eduardo Teixeira, Professor of Mathematics, Department of Mathematics, University of Central Florida, USA

  May 15, 16:15 - 17:30

  B1 L4 R4102

  This mini-course presents a powerful perturbative framework tailored to tackle critical regularity issues in nonlinear diffusion PDE.
• 

  The Signed Translation Transformed Depth: Order, Quantiles, Spread, Skewness, and Quantile Regression for Multivariate Functional Data

  Emmanuel Ambriz, Postdoctoral Research Fellow, Statistics

  May 15, 12:00 - 13:00

  B9 L2 R2325

  multivariate analysis

  This seminar introduces the Signed Translation transformed Depth (STtD), a novel interpretable ordering method for multivariate functional data, which enables enhanced distributional analysis, the definition of descriptive tools, and a flexible vine copula-based quantile regression framework.
• 

  Perturbation Methods in PDE: from Heuristics to Breakthroughs in Regularity Theory - Part 2

  Dr. Eduardo Teixeira, Professor of Mathematics, Department of Mathematics, University of Central Florida, USA

  May 14, 16:15 - 17:30

  B1 L3 R3119

  This mini-course presents a powerful perturbative framework tailored to tackle critical regularity issues in nonlinear diffusion PDE.
• 

  Nonlocal Degenerate PDEs: Bridging Free Boundary Problems and Critical-point Models

  Eduardo Teixeira, Graduate Director & Professor, Department of Mathematics, University of Central Florida

  May 13, 16:00 - 17:00

  Building 1, Level 3, Room 3426

  nonlocal degeneracies local extrema models regularity estimates

  This talk explores a new class of PDEs featuring nonlocal degeneracies. Our framework unifies two classical scenarios — free boundary problems and critical-point degenerate PDEs; both are recast as local extrema models in our formulation.
• 

  Topological Phenomena in Artificial Materials

  Xiujuan Zhang, Associate Professor, School of Engineering and Applied Sciences, Department of Material Science and Engineering, Nanjing University

  May 12, 10:00 - 11:00

  B1 L3 R3119

  topological phenomena artificial materials

  This talk will cover the design and realization of novel topological states and effects in acoustic artificial materials, focusing on higher-order topological states, non-Hermitian physics, and acoustic spin- and orbital angular momentum-related topological phenomena.
• 

  Perturbation Methods in PDE: from Heuristics to Breakthroughs in Regularity Theory - Part 1

  Dr. Eduardo Teixeira, Professor of Mathematics, Department of Mathematics, University of Central Florida, USA

  May 11, 16:15 - 17:30

  B1 L3 R3119

  This mini-course presents a powerful perturbative framework tailored to tackle critical regularity issues in nonlinear diffusion PDE.

May 10 - May 4, 2025

• 

  Vecchia Approximations of Gaussian Processes on GPUs for Scalable Spatial Modeling and Computer Model Emulation

  Qilong Pan, Ph.D. Student, Statistics

  May 8, 12:00 - 13:00

  B9 L2 R2325

  machine learning Geospatial Data GPU Computing

  This seminar introduces GPU-accelerated Vecchia approximations to overcome Gaussian Process computational limits, enabling scalable applications for large geospatial datasets and high-dimensional computer model emulations.
• 

  Current and Future Challenges and Solutions in AI & HPC System and Thermal Management

  Dr. Gamal Refai-Ahmed, Senior Fellow & Chief Architect, AMD Member of U.S. National Academy of Engineering Life Fellow, Canadian Academy of Engineering Fellow, Engineering Institute of Canada Fellow & Distinguished Lecturer, IEEE Life Fellow, ASME

  May 6, 13:00 - 17:00

  B4 L5 R5209

  Led by expert Dr. Gamal Refai Ahmed, this course explores innovative thermal management and packaging solutions for AI and HPC systems, addressing current and future challenges with cutting-edge techniques and next-generation design principles.
• 

  First-Order Mean-Field Games with Entry-Exit Flow Constraints and Contact-Set Conditions

  AbdulRahman Alharbi, Ph.D. Student, Applied Mathematics and Computational Science

  May 4, 12:30 - 14:30

  B9 L4 R4225; Zoom Meeting 95071981979

  mean-field models free boundary problems

  This dissertation develops and rigorously analyzes first-order mean-field game models incorporating novel mixed boundary conditions to realistically represent population dynamics in bounded domains by eliminating unrealistic entry phenomena that are typically induced by standard Dirichlet conditions.

May 3 - Apr 27, 2025

• 

  Retraction Maps, Feedback Linearization and Nonholonomic Integrators

  Ravi Banavar, Professor, Systems and Control Engineering, IIT Bombay

  May 1, 12:00 - 13:00

  B9 L2 R2325

  This talk will introduce the utility of retraction maps on Riemannian manifolds to two applications in applied mechanics and control.

Apr 26 - Apr 20, 2025

• 

  Linear Solvers for Large-Scale Bayesian Modeling

  Lisa Gaedke-Merzhäuser, Postdoctoral Research Fellow, Statistics

  Apr 24, 12:00 - 13:00

  B9 L2 R2325

  In this talk we explore what it means to perform Bayesian inference and introduce the methodology of integrated nested Laplace approximations (INLA).

Apr 19 - Apr 13, 2025

• 

  Homogenization and Wave Functional Materials

  Ying Wu, Program Chair, Applied Mathematics and Computational Science

  Apr 17, 12:00 - 13:00

  B9 L2 R2325

  Wave functional materials Homogenization

  In this talk, I will give a brief review on the progress from on homogenization and its impact on achieving unusual wave functional materials, such as zero-index materials in which the effective refractive index is vanishing.

Apr 12 - Apr 6, 2025

• 

  The Earliest Arrival of Quantum Advantage

  David Keyes, Professor, Applied Mathematics and Computational Science

  Apr 10, 12:00 - 13:00

  B9 L2 R2325

  quantum computing quantum processing unit supercomputers

  This talk outlines a "quantum first" strategy for future supercomputers, integrating QPUs, GPUs, and CPUs to optimize energy efficiency and accelerate scientific computing while addressing the current challenges and projecting the maturation of quantum computing by leveraging classical supercomputing advancements.
• 

  Helmholtz–Korteweg Equations: Modeling, Analysis, and Applications

  Umberto Zerbinati

  Apr 8, 15:00 - 16:00

  B1 L4 R4102

  Korteweg fluids Helmholtz–Korteweg equation

  This talk presents a derivation and analysis of Helmholtz-Korteweg equations, extended to nematic-Korteweg fluids, including well-posedness proofs, numerical methods, and predictions for wave propagation and scattering phenomena with experimental implications.
• 

  Workshop on Analysis and PDEs

  Apr 7, 08:00 - Apr 10, 17:00

  B3, L5, R5220

  PDEs Analysis of PDE's fluid dynamics free boundary problems energy minimization Mathematical Biology

  Join us for the Workshop on PDEs to explore advances, challenges and methods in the analysis and computation of partial differential equations.

Mar 29 - Mar 23, 2025

• 

  Some Contributions to Particle and Unbiased Simulation Methods

  Elsiddig Awadelkarim Elsiddig, Postdoctoral Research Fellow, Statistics

  Mar 26, 10:00 - 12:00

  B3 L5 R5209; Zoom Meeting 95303621356

  Bayesian Statistics machine learning

  This thesis develops unbiased estimators and efficient particle filters, leveraging Feynman-Kac formulae and multilevel techniques, to address challenges in parameter estimation, filtering, invariant measure approximation, and stationary distribution computation for partially observed diffusion processes and McKean-Vlasov stochastic differential equations.
• 

  Dimension Liftings for Quantum Computation of Partial Differential Equations and Related Problems

  Shi Jin, Director of Institute of Natural Sciences, Chair Professor of Mathematics, Shanghai Jiao Tong University

  Mar 25, 16:00 - 17:00

  B1 L3 R3119

  PDEs Schrödinger equation quantum simulation quantum computing

  This lecture presents a systematic framework for developing quantum simulation algorithms for general differential equations, using dimension lifting to transform nonlinear and time-dependent problems into forms solvable by simulating the Schrödinger equation.

Mar 22 - Mar 16, 2025

• 

  Extensions of Dynamic Programming for Combinatorial Optimization and Data Mining

  Mikhail Moshkov, Professor, Applied Mathematics and Computational Science

  Mar 20, 12:00 - 13:00

  B9, L2, R2325

  Dynamic programming is an efficient technique to solve optimization problems. It is based on decomposing the initial problem into simpler ones and solving these sub-problems beginning from the simplest ones. A conventional dynamic programming algorithm returns an optimal object from a given set of objects.

Mar 15 - Mar 9, 2025

• 

  On the Use of "Conventional" Unconstrained Minimization Solvers for Training Regression Problems in Scientific Machine Learning

  Stefano Zampini, Senior Research Scientist, Hierarchical Computations on Manycore Architectures

  Mar 13, 12:00 - 13:00

  B9 L2 R2325

  petsc PETScML machine learning

  This talk introduces PETScML, a framework leveraging traditional second-order optimization solvers for use within scientific machine learning, demonstrating improved generalization capabilities over gradient-based methods routinely adopted in deep learning.
• 

  Viscous Thin Films Driven by Thermal Noise

  Manuel V. Gnann, Assistant Professor, Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, Netherlands

  Mar 11, 16:00 - 17:00

  B1 L3 R3119

  This talk explores the mathematical analysis of stochastic thin-film equations, focusing on the existence of solutions for nonlinear noise and measure-valued solutions, highlighting the importance of the Stratonovich formulation for accurate physical modelling of capillary-driven viscous thin films at microscopic scales.
• 

  Chase and Escape from Sticks to Groups: Effects of Noise and Delay

  Toru Ohira, Professor, Graduate School of Mathematics Division of Mathematics Advanced Topology, Nagoya University, Japan

  Mar 10, 16:00 - 17:00

  B1 L3 R3119

  This talk explores the "group chase and escape" model, an extension of the classic chase and escape problem, demonstrating emergent complex behaviours like emergent aggregate formations, noise-enhanced catching, and distance-dependent movements delays, and further examines the impact of probabilistically converting caught escapees into chasers, adding another layer of dynamical complexity to the system.

Mar 8 - Mar 2, 2025

• 

  The Ball-Proximal (=“Broximal”) Point Method: a New Algorithm, Convergence Theory, and Applications

  Kaja Gruntkowska, PhD student, Statistics, KAUST

  Mar 6, 12:00 - 13:00

  B9, L2, R2325

  Non-smooth and non-convex global optimization poses significant challenges across various appli- cations, where standard gradient-based methods often struggle. We propose the Ball-Proximal Point Method, Broximal Point Method, or Ball Point Method (BPM) for short – a novel algorith- mic framework inspired by the classical Proximal Point Method (PPM) (Rockafellar, 1976), which, as we show, sheds new light on several founda- tional optimization paradigms and phenomena, including non-convex and non-smooth optimiza- tion, acceleration, smoothing, adaptive stepsize selection, and trust-region methods.
• 

  Enhance Neural Network Training for Complex PDEs: Algorithms, Architectures, and Integration with Traditional Methods

  Dr. Yahong Yang, Postdoctoral Scholar, Pennsylvania State University, USA

  Mar 2, 12:00 - 13:00

  Auditorium between Building 2 - 3, L0 R0215

  This talk explores recent research on improving neural network methods for solving partial differential equations (PDEs) by addressing training challenges through novel algorithms, architectures, and integration with traditional numerical techniques.

Mar 1 - Feb 23, 2025

• 

  Ringmaster ASGD: The First Asynchronous SGD with Optimal Time Complexity

  Arto Maranjyan, Ph.D. Student, Computer Science

  Feb 27, 12:00 - 13:00

  B9, L2, R2325

  Asynchronous Stochastic Gradient Descent (Asynchronous SGD) is a cornerstone method for parallelizing learning in distributed machine learning. However, its performance suffers under arbitrarily heterogeneous computation times across workers, leading to suboptimal time complexity and inefficiency as the number of workers scales.
• 

  Adapting Foundation Models: From Reinforcement Learning to Multivariate Time Series Forecasting

  Abdelhakim Benechehab, Ph.D. Student, EURECOM, France

  Feb 24, 12:00 - 13:00

  B4 L5 R5220

  This talk explores adapting foundation models for reinforcement learning and multivariate time series forecasting, introducing DICL for handling multivariate data in RL and AdaPTS for probabilistic time series forecasting, both showing significant performance improvements over baseline methods.
• 

  Theoretical Foundations of Efficient and Scalable Graph Learning

  Dr. Ziang Chen, Instructor, Applied Mathematics, MIT, USA

  Feb 24, 12:00 - 13:00

  Auditorium between Building 2 - 3, L0 R0215

  Graph Neural Networks

  This talk addresses key challenges in graph neural networks (GNNs), including expressive power and model depth, by theoretically analyzing and providing solutions in three areas: universal approximation of linear program properties, strong expressive power of subgraph GNNs on graphs with bounded cycles, and the role of residual connections in mitigating oversmoothing and enabling deeper models.

Feb 22 - Feb 16, 2025

• 

  Derivative-Free Methods in Nonconvex Optimization with and Without Noise

  Boris Mordukhovich, Distinguished University Professor, Wayne State University

  Feb 20, 12:00 - 13:00

  B9, L2, R2325

  This talk addresses the study of nonconvex derivative-free optimization problems, where only information of either smooth objective functions or their noisy approximations is available. General derivative-free methods are proposed for minimizing differentiable (not necessarily convex) functions with globally Lipschitz continuous gradients, where the accuracy of approximate gradients is interacting with stepsizes and exact gradient values.