GRAMSIA - Graphical Models, Statistical Inference, and Algorithms

Explore advanced mathematical foundations of graphical models and statistical inference through this comprehensive workshop featuring leading researchers from MIT, Brown University, and Carnegie Mellon. Delve into cutting-edge topics including mean-field approximations for high-dimensional Bayesian regression, spectral statistics for sparse random graphs, and modern perspectives on reconstruction bounds for tree structures. Examine algorithmic approaches to planted clique problems, broadcasting on trees, and stochastic block models while investigating the theoretical underpinnings of machine learning through Gaussian data analysis. Discover recent developments in random graph matching, counterfactual inference with unobserved confounding, and distribution compression techniques. Learn about neighborhood interference with low-order interactions, implicit regularization through uniform convergence, and margin-maximization via game-theoretic approaches. Gain insights into algorithmic decorrelation methods, augmented IPW estimators in high dimensions, and optimization strategies that bridge theoretical mathematics with practical machine learning applications.