Saved in:
Bibliographic Details
Main Authors: Hua, Andong, Gu, Jindong, Xue, Zhiyu, Carlini, Nicholas, Wong, Eric, Qin, Yao
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2312.05716
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911819860279296
author Hua, Andong
Gu, Jindong
Xue, Zhiyu
Carlini, Nicholas
Wong, Eric
Qin, Yao
author_facet Hua, Andong
Gu, Jindong
Xue, Zhiyu
Carlini, Nicholas
Wong, Eric
Qin, Yao
contents With the prevalence of the Pretraining-Finetuning paradigm in transfer learning, the robustness of downstream tasks has become a critical concern. In this work, we delve into adversarial robustness in transfer learning and reveal the critical role of initialization, including both the pretrained model and the linear head. First, we discover the necessity of an adversarially robust pretrained model. Specifically, we reveal that with a standard pretrained model, Parameter-Efficient Finetuning (PEFT) methods either fail to be adversarially robust or continue to exhibit significantly degraded adversarial robustness on downstream tasks, even with adversarial training during finetuning. Leveraging a robust pretrained model, surprisingly, we observe that a simple linear probing can outperform full finetuning and other PEFT methods with random initialization on certain datasets. We further identify that linear probing excels in preserving robustness from the robust pretraining. Based on this, we propose Robust Linear Initialization (RoLI) for adversarial finetuning, which initializes the linear head with the weights obtained by adversarial linear probing to maximally inherit the robustness from pretraining. Across five different image classification datasets, we demonstrate the effectiveness of RoLI and achieve new state-of-the-art results. Our code is available at \url{https://github.com/DongXzz/RoLI}.
format Preprint
id arxiv_https___arxiv_org_abs_2312_05716
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Initialization Matters for Adversarial Transfer Learning
Hua, Andong
Gu, Jindong
Xue, Zhiyu
Carlini, Nicholas
Wong, Eric
Qin, Yao
Computer Vision and Pattern Recognition
With the prevalence of the Pretraining-Finetuning paradigm in transfer learning, the robustness of downstream tasks has become a critical concern. In this work, we delve into adversarial robustness in transfer learning and reveal the critical role of initialization, including both the pretrained model and the linear head. First, we discover the necessity of an adversarially robust pretrained model. Specifically, we reveal that with a standard pretrained model, Parameter-Efficient Finetuning (PEFT) methods either fail to be adversarially robust or continue to exhibit significantly degraded adversarial robustness on downstream tasks, even with adversarial training during finetuning. Leveraging a robust pretrained model, surprisingly, we observe that a simple linear probing can outperform full finetuning and other PEFT methods with random initialization on certain datasets. We further identify that linear probing excels in preserving robustness from the robust pretraining. Based on this, we propose Robust Linear Initialization (RoLI) for adversarial finetuning, which initializes the linear head with the weights obtained by adversarial linear probing to maximally inherit the robustness from pretraining. Across five different image classification datasets, we demonstrate the effectiveness of RoLI and achieve new state-of-the-art results. Our code is available at \url{https://github.com/DongXzz/RoLI}.
title Initialization Matters for Adversarial Transfer Learning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2312.05716