Huishuai Zhang

Indiscriminate Poisoning Attacks Are Shortcuts.

2021 · arXiv (Cornell University)

Indiscriminate data poisoning attacks, which add imperceptible perturbations to training data to maximize the test error of trained models, have become a trendy topic because they are thought to be capable of preventing unauthorized use …

Differentially Private Fine-tuning of Language Models

2024 · Journal of Privacy and Confidentiality

We give simpler, sparser, and faster algorithms for differentially private fine-tuning of large-scale pre-trained language models, which achieve the state-of-the-art privacy versus utility tradeoffs on many standard NLP tasks. We propose a meta-framework for this …