Unveiling Vulnerability

Research into "unveiling vulnerability" focuses on identifying and mitigating weaknesses in various machine learning models and data structures, particularly concerning privacy, security, and fairness. Current efforts concentrate on analyzing vulnerabilities in graph data structures, large language models (LLMs), neural networks (including those employing self-attention and contrastive learning), and text-to-image models, often employing adversarial attacks to probe model weaknesses. These findings are crucial for enhancing the trustworthiness and robustness of AI systems across diverse applications, ranging from recommender systems to chip design and abusive language detection. The ultimate goal is to develop more secure, reliable, and ethically sound AI technologies.

Papers

July 26, 2024

Unveiling Privacy Vulnerabilities: Investigating the Role of Structure in Graph Data
Hanyang Yuan, Jiarong Xu, Cong Wang, Ziqi Yang, Chunping Wang, Keting Yin, Yang Yang
Integral Role Inner Structure Graph Data PRIvacy Leakage Privacy Utility Attribute Inference Attack Unveiling Vulnerability

May 11, 2024

LLMs and the Future of Chip Design: Unveiling Security Risks and Building Trust
Zeng Wang, Lilas Alrahis, Likhitha Mankali, Johann Knechtel, Ozgur Sinanoglu
Medical LLM Future Reasoning Building Trust Chip Design Electronic Design Automation Hardware Description Language Circuit Model Unveiling Vulnerability

March 25, 2024

Revealing Vulnerabilities of Neural Networks in Parameter Learning and Defense Against Explanation-Aware Backdoors
Md Abdul Kadir, GowthamKrishna Addluri, Daniel Sonntag
Neural Network Explainable Artificial Intelligence XAI Method xAI Community Parameter Learning Unveiling Vulnerability

February 26, 2024

Unveiling Vulnerability of Self-Attention
Khai Jiet Liong, Hongqiu Wu, Hai Zhao
Language Model Adversarial Training Self Attention Level Perturbation Unveiling Vulnerability Textual Attack

February 16, 2024

ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages
Junjie Ye, Sixian Li, Guanyu Li, Caishuang Huang, Songyang Gao, Yilong Wu, Qi Zhang, Tao Gui, Xuanjing Huang
Large Language Model BRIO Tool Closed Source Large Language Model Tool Learning Multiple Stage Unveiling Vulnerability

January 16, 2024

Revealing Vulnerabilities in Stable Diffusion via Targeted Attacks
Chenyu Zhang, Lanjun Wang, Anan Liu
Adversarial Attack Stable Diffusion Adversarial Prompt Targeted Attack Unveiling Vulnerability

November 30, 2023

Unveiling Vulnerabilities of Contrastive Recommender Systems to Poisoning Attacks
Zongwei Wang, Junliang Yu, Min Gao, Hongzhi Yin, Bin Cui, Shazia Sadiq
Contrastive Learning Recommender System Poisoning Attack Recommendation Accuracy Unveiling Vulnerability Hypergraph Contrastive Feature Dispersion

November 15, 2023

Beyond Detection: Unveiling Fairness Vulnerabilities in Abusive Language Models
Yueqing Liang, Lu Cheng, Ali Payani, Kai Shu
Data Detection Backdoor Attack Abusive Language Detection Effective Model Fair Adversarial Robust Fairness Unveiling Vulnerability Abusive Language Classifier

November 7, 2023

Unveiling Safety Vulnerabilities of Large Language Models
George Kour, Marcel Zalmanovici, Naama Zwerdling, Esther Goldbraich, Ora Nova Fandina, Ateret Anaby-Tavor, Orna Raz, Eitan Farchi
Large Language Model Adversarial Example Semantic Region Unveiling Vulnerability

July 21, 2023

Unveiling Vulnerabilities in Interpretable Deep Learning Systems with Query-Efficient Black-box Attacks
Eldor Abdukhamidov, Mohammed Abuhamad, Simon S. Woo, Eric Chan-Tin, Tamer Abuhmed
Adversarial Attack Adversarial Example New Attack Attack Success Rate Interpretable Deep Learning Unveiling Vulnerability

Unveiling Vulnerability

Papers

Unveiling Privacy Vulnerabilities: Investigating the Role of Structure in Graph Data

LLMs and the Future of Chip Design: Unveiling Security Risks and Building Trust

Revealing Vulnerabilities of Neural Networks in Parameter Learning and Defense Against Explanation-Aware Backdoors

Unveiling Vulnerability of Self-Attention

ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages

Revealing Vulnerabilities in Stable Diffusion via Targeted Attacks

Unveiling Vulnerabilities of Contrastive Recommender Systems to Poisoning Attacks

Beyond Detection: Unveiling Fairness Vulnerabilities in Abusive Language Models

Unveiling Safety Vulnerabilities of Large Language Models

Unveiling Vulnerabilities in Interpretable Deep Learning Systems with Query-Efficient Black-box Attacks