12 papers
Sensitivity of SDE Solutions to Perturbations of the Diffusion and Drift
Jeremiah Birrell
We develop a method for bounding the sensitivity of solutions to stochastic differential equations (SDEs) to changes in the drift, , and diffusion, , by using a combination…
Information Theoretic Adversarial Training of Large Language Models
Yiwei Zhang, Jeremiah Birrell, Reza Ebrahimi +3
Large language models (LLMs) remain vulnerable to adversarial prompting despite advances in alignment and safety, often exhibiting harmful behaviors under novel attack strategies.…
Statistical Guarantees for Distributionally Robust Optimization with Optimal Transport and OT-Regularized Divergences
Jeremiah Birrell, Xiaoxi Shen
We study finite-sample statistical performance guarantees for distributionally robust optimization (DRO) with optimal transport (OT) and OT-regularized divergence model neighborhoo…
Concentration Inequalities for Stochastic Optimization of Unbounded Objective Functions with Application to Denoising Score Matching
Jeremiah Birrell
We derive novel concentration inequalities that bound the statistical error for a large class of stochastic optimization problems, focusing on the case of unbounded objective funct…
Local Differential Privacy for Federated Learning with Fixed Memory Usage and Per-Client Privacy
Rouzbeh Behnia, Jeremiah Birrell, Arman Riasi +3
Federated learning (FL) enables organizations to collaboratively train models without sharing their datasets. Despite this advantage, recent studies show that both client updates a…
Concentration Inequalities and UQ Bounds for Hypocoercive MCMC Samplers
Jeremiah Birrell, Luc Rey-Bellet
In this work we provide performance guarantees for hypocoercive non-reversible MCMC samplers with invariant measure ; our results apply in particular to the Langevin eq…