GradMask: Gradient-Guided Token Masking for Textual Adversarial Example Detection
We present a simple model-agnostic textual adversarial example detection scheme called GradMask. It uses gradient signals to detect adversarially perturbed tokens in an input sequence and occludes such tokens by a masking process. GradMask provides several advantages over existing methods including improved detection performance and a weak interpretation of its decision. Extensive evaluations on widely adopted natural language processing benchmark datasets demonstrate the efficiency and effectiveness of GradMask
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
GradMask: Effective Fine-tuning on Large-scale Pretrained Language Models via Gradient Masking
Pretrained language models have dominated a variety of NLP tasks. However, fine-tuning large pretrained models on downstream tasks tend to achieve degenerated and unstable results, especially when there are only a limite…
Gradient-Attention Guided Dual-Masking Synergetic Framework for Robust Text-based Person Retrieval
Although Contrastive Language-Image Pre-training (CLIP) exhibits strong performance across diverse vision tasks, its application to person representation learning faces two critical challenges: (i) the scarcity of large-…
Representation LearningContrastive LearningPerson RetrievalGradMask: Reduce Overfitting by Regularizing Saliency
With too few samples or too many model parameters, overfitting can inhibit the ability to generalise predictions to new data. Within medical imaging, this can occur when features are incorrectly assigned importance such …
Lesion SegmentationAttribution-Guided Masking for Robust Cross-Domain Sentiment Classification
While pre-trained Transformer models achieve high accuracy on in-domain sentiment classification, they frequently experience severe performance degradation when transferring to out-of-domain data. We hypothesize that thi…
Localize and Neutralize: Gradient-guided Token Suppression against Visual Prompt Injection Attack
Adversarial images pose a severe security threat to multimodal large language models through prompt injection. Existing defenses largely lack a principled understanding of the underlying mechanisms and struggle to balanc…
Adversarial Attack