AdCorDA: Classifier Refinement via Adversarial Correction and Domain Adaptation

Lulan Shen, Ali Edalati, Xiangyu Li, Shilei Lin, Brett H. Meyer, Warren J. Gross, James J. Clark

Keywords: adversarial correction, domain adaptation, curriculum learning, adversarial attacks

Abstract: This paper describes a simple yet effective technique for refining a pretrained classifier network. The proposed AdCorDA method consists of two stages - adversarial correction followed by domain adaptation. Adversarial correction uses adversarial attacks to correct misclassified training-set classifications. The incorrectly classified samples of the training set are removed and replaced with the adversarially corrected samples to form a new training set, and then, in the second stage, domain adaptation is performed back to the original training set. Extensive experimental validations show significant accuracy boosts of over 5% on the CIFAR-100 dataset and 1% on the CINIC-10 dataset. The technique can be straightforwardly applied to the refinement of weight-quantized neural networks, where experiments show substantial enhancement in performance over the baseline. The adversarial correction technique also results in enhanced robustness to adversarial attacks.

Primary Area: transfer learning, meta learning, and lifelong learning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 4525