paper-with-me

홈 › Papers

Improved OOD Generalization via Adversarial Training and Pre-training

2021-05-24 · Mingyang Yi, Lu Hou, Jiacheng Sun, Lifeng Shang, Xin Jiang, Qun Liu, Zhi-Ming Ma

Recently, learning a model that generalizes well on out-of-distribution (OOD) data has attracted great attention in the machine learning community. In this paper, after defining OOD generalization via Wasserstein distance, we theoretically show that a model robust to input perturbation generalizes well on OOD data. Inspired by previous findings that adversarial training helps improve input-robustness, we theoretically show that adversarially trained models have converged excess risk on OOD data, and empirically verify it on both image classification and natural language understanding tasks. Besides, in the paradigm of first pre-training and then fine-tuning, we theoretically show that a pre-trained model that is more robust to input perturbation provides a better initialization for generalization on downstream OOD data. Empirically, after fine-tuning, this better-initialized model from adversarial pre-training also has better OOD generalization.

📄 PDF Abstract BibTeX arXiv:2105.11144

Code (0)

등록된 구현이 없습니다.

Tasks

image-classificationImage ClassificationNatural Language Understanding

Similar Papers 제목 키워드 기반

Stability and Generalization in Free Adversarial Training

2024-04-13 · Xiwei Cheng, Kexin Fu, Farzan Farnia

While adversarial training methods have significantly improved the robustness of deep neural networks against norm-bounded adversarial perturbations, the generalization gap between their performance on training and test …

Generalization Bounds

Impact of Adversarial Training on Robustness and Generalizability of Language Models

2022-11-10 · Enes Altinisik, Hassan Sajjad, Husrev Taha Sencar, Safa Messaoud 외

Adversarial training is widely acknowledged as the most effective defense against adversarial attacks. However, it is also well established that achieving both robustness and generalization in adversarially trained model…

Data Augmentation

CAT: Customized Adversarial Training for Improved Robustness

2020-02-17 · Minhao Cheng, Qi Lei, Pin-Yu Chen, Inderjit Dhillon 외

Adversarial training has become one of the most effective methods for improving robustness of neural networks. However, it often suffers from poor generalization on both clean and perturbed data. In this paper, we propos…

Improving the Generalization of Adversarial Training with Domain Adaptation

2018-10-01 · ICLR 2019 5 · Chuanbiao Song, Kun He, Li-Wei Wang, John E. Hopcroft

By injecting adversarial examples into training data, adversarial training is promising for improving the robustness of deep learning models. However, most existing adversarial training approaches are based on a specific…

Adversarial AttackDomain Adaptation

Generating Less Certain Adversarial Examples Improves Robust Generalization

2023-10-06 · Minxing Zhang, Michael Backes, Xiao Zhang

This paper revisits the robust overfitting phenomenon of adversarial training. Observing that models with better robust generalization performance are less certain in predicting adversarially generated training inputs, w…