🤖 AI Summary
Transformer-based text classifiers (e.g., BERT, RoBERTa) are vulnerable to adversarial examples, and existing defenses lack interpretability and traceability. To address this, we propose the first holistic adversarial example lifecycle management framework that jointly integrates model interpretability and human expert feedback. Our method enables transparent detection and attribution via attention maps and integrated gradients; repairs perturbed tokens under semantic consistency constraints using pretrained word embedding substitution and model-guided optimization; and incorporates a human-in-the-loop review mechanism to ensure trustworthy transformation decisions. Evaluated on multiple BERT and RoBERTa benchmarks, our approach achieves >92% adversarial detection rate and 89% harmless transformation rate, significantly enhancing model robustness and auditability of security analysis.
📝 Abstract
Transformer-based text classifiers like BERT, Roberta, T5, and GPT-3 have shown impressive performance in NLP. However, their vulnerability to adversarial examples poses a security risk. Existing defense methods lack interpretability, making it hard to understand adversarial classifications and identify model vulnerabilities. To address this, we propose the Interpretability and Transparency-Driven Detection and Transformation (IT-DT) framework. It focuses on interpretability and transparency in detecting and transforming textual adversarial examples. IT-DT utilizes techniques like attention maps, integrated gradients, and model feedback for interpretability during detection. This helps identify salient features and perturbed words contributing to adversarial classifications. In the transformation phase, IT-DT uses pre-trained embeddings and model feedback to generate optimal replacements for perturbed words. By finding suitable substitutions, we aim to convert adversarial examples into non-adversarial counterparts that align with the model's intended behavior while preserving the text's meaning. Transparency is emphasized through human expert involvement. Experts review and provide feedback on detection and transformation results, enhancing decision-making, especially in complex scenarios. The framework generates insights and threat intelligence empowering analysts to identify vulnerabilities and improve model robustness. Comprehensive experiments demonstrate the effectiveness of IT-DT in detecting and transforming adversarial examples. The approach enhances interpretability, provides transparency, and enables accurate identification and successful transformation of adversarial inputs. By combining technical analysis and human expertise, IT-DT significantly improves the resilience and trustworthiness of transformer-based text classifiers against adversarial attacks.
