White Box

"White-box" research in machine learning focuses on analyzing and manipulating models with complete access to their internal parameters and workings, primarily to assess vulnerabilities and improve security. Current research emphasizes adversarial attacks (e.g., poisoning training data, crafting adversarial examples) and defenses against these attacks, often targeting specific model architectures like transformers and graph neural networks, as well as exploring techniques like watermarking for intellectual property protection. This research is crucial for building more robust and trustworthy AI systems, impacting the security of various applications from autonomous vehicles to large language models and mitigating risks associated with data privacy and model integrity.

Papers

November 12, 2024

Zer0-Jack: A Memory-efficient Gradient-based Jailbreaking Method for Black-box Multi-modal Large Language Models
Tiejin Chen, Kaishen Wang, Hua Wei
Jailbreak Attack Black Box Model Multi Modal Large Language Model White Box Malicious Traffic Memory effIcieNt

November 11, 2024

On Active Privacy Auditing in Supervised Fine-tuning for White-Box Language Models
Qian Sun, Hanpeng Wu, Xi Sheryl Zhang
Language Model Supervised Fine Tuning White Box Empirical Privacy Private Fine Tuning Box Membership Inference Attack

October 2, 2024

Signal Adversarial Examples Generation for Signal Detection Network via White-Box Attack
Dongyang Li, Linyuan Wang, Guangwei Xiong, Bin Yan, Dekui Ma, Jinxian Peng
Adversarial Attack Adversarial Example White Box Detection Network

October 1, 2024

Timber! Poisoning Decision Trees
Stefano Calzavara, Lorenzo Cazzaro, Massimo Vettori
Decision Tree Poisoning Attack Tree Specie White Box Model Retraining Greedy Approach

September 16, 2024

FreeMark: A Non-Invasive White-Box Watermarking for Deep Neural Networks
Yuzhang Chen, Jiangnan Zhu, Yujie Gu, Minoru Kuribayashi, Kouichi Sakurai
Deep Neural Network White Box Watermark Removal Attack DNN Watermarking

September 10, 2024

On the Weaknesses of Backdoor-based Model Watermarking: An Information-theoretic Perspective
Aoting Hu, Yanzhi Chen, Renjie Xie, Adrian Weller
Information Theoretic Watermarking Method White Box Model Watermarking Physical Weakness Watermark Removal Attack

September 3, 2024

Exploiting the Vulnerability of Large Language Models via Defense-Aware Architectural Backdoor
Abdullah Arafat Miah, Yu Bi
Backdoor Attack White Box Latent Vulnerability

August 27, 2024

Adversarial Attacks and Defenses in Multivariate Time-Series Forecasting for Smart and Connected Infrastructures
Pooja Krishan, Rohan Mohapatra, Saptarshi Sengupta
Adversarial Attack Deep Learning Model Adversarial Training Time Series Forecasting Adversarial DEfense White Box Smart Infrastructure

July 16, 2024

Are Linear Regression Models White Box and Interpretable?
Ahmed M Salih, Yuhe Wang
Inherent Interpretability High Explainability Linear Regression Interpretable Way White Box

July 15, 2024

Provable Robustness of (Graph) Neural Networks Against Data Poisoning and Backdoor Attacks
Lukas Gosch, Mahalakshmi Sabanayagam, Debarghya Ghoshdastidar, Stephan Günnemann
Neural Network Graph Neural Network Graph Drawing Backdoor Attack White Box Data Poisoning Worst Case Robustness Adversarial Change

July 4, 2024

TrackPGD: A White-box Attack using Binary Masks against Robust Transformer Trackers
Fatemeh Nourilenjan Nokabadi, Yann Batiste Pequignot, Jean-Francois Lalonde, Christian Gagné
Adversarial Robustness White Box Binary Mask Transformer Tracker White Box Adversarial Attack

June 17, 2024

Do Parameters Reveal More than Loss for Membership Inference?
Anshuman Suri, Xiao Zhang, David Evans
Membership Inference Attack Many Parameter White Box Membership Inference Interest Loss Membership Privacy Hessian Inverse

May 6, 2024

On Adversarial Examples for Text Classification by Perturbing Latent Representations
Korn Sooksatra, Bikram Khanal, Pablo Rivas
Deep Learning Adversarial Example Text Classification White Box Hidden Representation Perturbation

May 4, 2024

Assessing Adversarial Robustness of Large Language Models: An Empirical Study
Zeyu Yang, Zhao Meng, Xiaochen Zheng, Roger Wattenhofer
Adversarial Attack Adversarial Robustness Adversarial Perturbation Empirical Study White Box LLM Robustness

April 6, 2024

PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics
Derui Zhu, Dingfan Chen, Qing Li, Zongxiong Chen, Lei Ma, Jens Grossklags, Mario Fritz
Large Language Model White Box Transition Dynamic Tractable Probabilistic Model Hallucination Annotator

April 3, 2024

Masked Completion via Structured Diffusion with White-Box Transformers
Druv Pai, Ziyang Wu, Sam Buchanan, Yaodong Yu, Yi Ma
Supervised Autoencoder Deep Network White Box Autoencoder Architecture Completion Method Structured Diffusion

March 13, 2024

RAF-GI: Towards Robust, Accurate and Fast-Convergent Gradient Inversion Attack in Federated Learning
Can Liu, Jin Wang, Dongyang Yu
White Box Inversion Performance Gradient Inversion Attack Gradient Inversion

March 10, 2024

Attacking Transformers with Feature Diversity Adversarial Perturbation
Chenxing Gao, Hang Zhou, Junqing Yu, YuTeng Ye, Jiale Cai, Junle Wang, Wei Yang
Vision Transformer Transformer Megatron Decepticons Adversarial Perturbation White Box

March 6, 2024

DeepEclipse: How to Break White-Box DNN-Watermarking Schemes
Alessandro Pegoraro, Carlotta Segna, Kavita Kumari, Ahmad-Reza Sadeghi
White Box Calysto Scheme DNN Watermarking

February 26, 2024

Pandora's White-Box: Precise Training Data Detection and Extraction in Large Language Models
Jeffrey G. Wang, Jason Wang, Marvin Li, Seth Neel
Membership Inference Attack Process Extraction White Box Privacy Attack Model Stealing PanDORA Dataset Detector Training

White Box

Papers

Zer0-Jack: A Memory-efficient Gradient-based Jailbreaking Method for Black-box Multi-modal Large Language Models

On Active Privacy Auditing in Supervised Fine-tuning for White-Box Language Models

Signal Adversarial Examples Generation for Signal Detection Network via White-Box Attack

Timber! Poisoning Decision Trees

FreeMark: A Non-Invasive White-Box Watermarking for Deep Neural Networks

On the Weaknesses of Backdoor-based Model Watermarking: An Information-theoretic Perspective

Exploiting the Vulnerability of Large Language Models via Defense-Aware Architectural Backdoor

Adversarial Attacks and Defenses in Multivariate Time-Series Forecasting for Smart and Connected Infrastructures

Are Linear Regression Models White Box and Interpretable?

Provable Robustness of (Graph) Neural Networks Against Data Poisoning and Backdoor Attacks

TrackPGD: A White-box Attack using Binary Masks against Robust Transformer Trackers

Do Parameters Reveal More than Loss for Membership Inference?

On Adversarial Examples for Text Classification by Perturbing Latent Representations

Assessing Adversarial Robustness of Large Language Models: An Empirical Study

PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics

Masked Completion via Structured Diffusion with White-Box Transformers

RAF-GI: Towards Robust, Accurate and Fast-Convergent Gradient Inversion Attack in Federated Learning

Attacking Transformers with Feature Diversity Adversarial Perturbation

DeepEclipse: How to Break White-Box DNN-Watermarking Schemes

Pandora's White-Box: Precise Training Data Detection and Extraction in Large Language Models