20 citations · 21 across the 4 of their papers we have counts for
5 papers
IF-GUIDE: Influence Function-Guided Detoxification of LLMs
Zachary Coalson, Juhan Bae, Nicholas Carlini +1
We study how training data contributes to the emergence of toxic behaviors in large language models. Most prior work on reducing model toxicity adopts reactive approaches, such as…
Privacy Backdoors: Enhancing Membership Inference through Poisoning Pre-trained Models
Yuxin Wen, Leo Marchyok, Sanghyun Hong +3
It is commonplace to produce application-specific models by fine-tuning large pre-trained models using a small bespoke dataset. The widespread availability of foundation model chec…
Certified Robustness to Clean-Label Poisoning Using Diffusion Denoising
Sanghyun Hong, Nicholas Carlini, Alexey Kurakin
We present a certified defense to clean-label poisoning attacks under -norm. These attacks work by injecting a small number of poisoning samples (e.g., 1%) that contain bou…
Publishing Efficient On-device Models Increases Adversarial Vulnerability
Sanghyun Hong, Nicholas Carlini, Alexey Kurakin
Recent increases in the computational demands of deep neural networks (DNNs) have sparked interest in efficient deep learning mechanisms, e.g., quantization or pruning. These mecha…
Handcrafted Backdoors in Deep Neural Networks
Sanghyun Hong, Nicholas Carlini, Alexey Kurakin
When machine learning training is outsourced to third parties, become practical as the third party who trains the model may act maliciously to inject hidden be…