Krishna Balasubramanian

Differentially Private Two-Stage Gradient Descent for Instrumental Variable Regression

Mirror Flow Matching with Heavy-Tailed Priors for Generative Modeling on Convex Domains

Dense Associative Memory on the Bures-Wasserstein Space

Restricted Spectral Gap Decomposition for Simulated Tempering Targeting Mixture Distributions

Dense Associative Memory with Epanechnikov Energy

Improved Finite-Particle Convergence Rates for Stein Variational Gradient Descent

Riemannian Proximal Sampler for High-accuracy Sampling on Manifolds

Transformers Handle Endogeneity in In-Context Linear Regression

Provable In-context Learning for Mixture of Linear Regressions using Transformers