Efficient Numerical Methods for Scalable Bayesian Inference
This talk will introduce and discuss the computational building blocks of Integrated Nested Laplace Approximations (INLA) and how these are implemented in software libraries such as R-INLA or DALIA.
Overview
Bayesian inference is used to model complex phenomena in fields such as epidemiology, finance, and environmental modeling, where uncertainty quantification is critical. In this talk, I will introduce the methodology of Integrated Nested Laplace Approximations (INLA), a powerful approach to performing Bayesian inference on latent Gaussian models. Beyond the conceptual framework, this talk will include techniques for efficiently handling sparse matrices, exploiting structured sparsity patterns, performing block-dense operations, and leveraging both CPU and GPU resources for efficient computation.
Presenters
Brief Biography
Lisa Gaedke-Merzhäuser is a Postdoctoral Fellow in the BayesComp group at KAUST, led by Prof. Håvard Rue. She holds a PhD in Computational Science from Università della Svizzera italiana (USI) in Lugano, Switzerland. Her research interests lie in fusing statistical learning techniques with methods from high-performance computing. She has been developing INLA_DIST, a distributed memory GPU-accelerated version of INLA for large-scale spatio-temporal models.
Related People
Related Researchers
Lisa Gaedke-Merzhäuser
- Postdoctoral Research Fellow, Statistics