paper-with-me

홈 › Papers

Balancing Robustness and Sensitivity using Feature Contrastive Learning

2021-05-19 · Seungyeon Kim, Daniel Glasner, Srikumar Ramalingam, Cho-Jui Hsieh, Kishore Papineni, Sanjiv Kumar

It is generally believed that robust training of extremely large networks is critical to their success in real-world applications. However, when taken to the extreme, methods that promote robustness can hurt the model's sensitivity to rare or underrepresented patterns. In this paper, we discuss this trade-off between sensitivity and robustness to natural (non-adversarial) perturbations by introducing two notions: contextual feature utility and contextual feature sensitivity. We propose Feature Contrastive Learning (FCL) that encourages a model to be more sensitive to the features that have higher contextual utility. Empirical results demonstrate that models trained with FCL achieve a better balance of robustness and sensitivity, leading to improved generalization in the presence of noise on both vision and NLP datasets.

📄 PDF Abstract BibTeX arXiv:2105.09394

Code (0)

등록된 구현이 없습니다.

Tasks

Contrastive LearningSensitivity

Methods 이 논문이 사용한 방법론

Contrastive Learning 설명 없음

Similar Papers 제목 키워드 기반

Balanced Adversarial Training: Balancing Tradeoffs Between Oversensitivity and Undersensitivity in NLP Models

2022-01-16 · ACL ARR January 2022 1 · Anonymous

Traditional (\emph{oversensitive}) adversarial examples involve finding a small perturbation that does not change an input's true label but confuses the classifier into outputting a different prediction. \emph{Undersensi…

Contrastive LearningNatural Language InferenceParaphrase Identification

Understanding and Improving Training-Free AI-Generated Image Detections with Vision Foundation Models

2024-11-28 · Chung-Ting Tsai, Ching-Yun Ko, I-Hsin Chung, Yu-Chiang Frank Wang 외

The rapid advancement of generative models has introduced serious risks, including deepfake techniques for facial synthesis and editing. Traditional approaches rely on training classifiers and enhancing generalizability …

DeepFake DetectionFace SwappingSensitivity

Pareto Adversarial Robustness: Balancing Spatial Robustness and Sensitivity-based Robustness

2021-11-03 · Ke Sun, Mingjie Li, Zhouchen Lin

Adversarial robustness, which primarily comprises sensitivity-based robustness and spatial robustness, plays an integral part in achieving robust generalization. In this paper, we endeavor to design strategies to achieve…

Adversarial RobustnessSensitivity

Adversarial Feature Alignment: Balancing Robustness and Accuracy in Deep Learning via Adversarial Training

2024-02-19 · Leo Hyun Park, JaeUk Kim, Myung Gyo Oh, Jaewoo Park 외

Deep learning models continue to advance in accuracy, yet they remain vulnerable to adversarial attacks, which often lead to the misclassification of adversarial examples. Adversarial training is used to mitigate this pr…

Contrastive LearningData Augmentation

Learning Time-Series Representations by Hierarchical Uniformity-Tolerance Latent Balancing

2025-10-02 · Amin Jalali, Milad Soltany, Michael Greenspan, Ali Etemad arxiv

We propose TimeHUT, a novel method for learning time-series representations by hierarchical uniformity-tolerance balancing of contrastive representations. Our method uses two distinct losses to learn strong representatio…

Temporal SequencesAnomaly Detection