Alexey Kurakin

Evaluating Privacy Risks of Parameter-Efficient Fine-Tuning

Certified Robustness to Clean-label Poisoning Using Diffusion Denoising

Harnessing large-language models to generate private synthetic text

DART: A Principled Approach to Adversarially Robust Unsupervised Domain Adaptation

Diffusion Denoising as a Certified Defense Against Clean-Label Poisoning Attacks