Membership Inference Attack

Membership inference attacks (MIAs) aim to determine if a specific data point was used to train a machine learning model, posing a significant privacy risk. Current research focuses on evaluating MIA effectiveness across various model architectures, including large language models (LLMs), diffusion models, and vision transformers, and exploring the impact of different training methods and data characteristics on attack success. The reliability and accuracy of MIAs themselves are under scrutiny, with some studies highlighting limitations and overestimation of their capabilities, particularly in realistic settings. Understanding the vulnerabilities and limitations of MIAs is crucial for developing effective privacy-preserving techniques and for responsibly deploying machine learning models.

Papers

September 29, 2024

September 28, 2024

Subject Data Auditing via Source Inference Attack in Cross-Silo Federated Learning
Jiaxin Li, Marco Arazzi, Antonino Nocera, Mauro Conti
Membership Inference Attack Data Leakage Inference Attack Cross Silo Federated Learning Data Auditing

September 22, 2024

Order of Magnitude Speedups for LLM Membership Inference
Martin Bertran, Rongting Zhang, Aaron Roth
Membership Inference Attack LLM Inference Privacy Risk Order Matter Linear Speedup Shadow Model

September 12, 2024

September 11, 2024

Context-Aware Membership Inference Attacks against Pre-trained Large Language Models
Hongyan Chang, Ali Shahin Shamsabadi, Kleomenis Katevas, Hamed Haddadi, Reza Shokri
Membership Inference Attack Pre Trained Large Language Model Token Sequence

September 5, 2024

Understanding Data Importance in Machine Learning Attacks: Does Valuable Data Pose Greater Harm?
Rui Wen, Michael Backes, Yang Zhang
Machine Learning Membership Inference Attack Potential Harm High Quality Data State of the Art Datasets

September 2, 2024

Membership Inference Attacks Against In-Context Learning
Rui Wen, Zheng Li, Michael Backes, Yang Zhang
Language Model Context Learning Membership Inference Attack Privacy Attack

August 16, 2024

MIA-Tuner: Adapting Large Language Models as Pre-training Text Detector
Wenjie Fu, Huandong Wang, Chen Gao, Guanghua Liu, Yong Li, Tao Jiang
Membership Inference Attack Text Detector

August 13, 2024

Membership Inference Attack Against Masked Image Modeling
Zheng Li, Xinlei He, Ning Yu, Yang Zhang
Membership Inference Attack Masked Image Modeling Mask Guided Image Encoder

August 9, 2024

Range Membership Inference Attacks
Jiashu Tao, Reza Shokri
Membership Inference Attack Private Data Empirical Privacy

August 2, 2024

Privacy-Preserving Split Learning with Vision Transformers using Patch-Wise Random and Noisy CutMix
Seungeun Oh, Sihun Baek, Jihong Park, Hyelin Nam, Praneeth Vepakomma, Ramesh Raskar, Mehdi Bennis, Seong-Lyun Kim
Vision Transformer Membership Inference Attack Split Learning Random Participation Differentially Private Private Training Cross Modal Manifold CutMix

August 1, 2024

Pathway to Secure and Trustworthy 6G for LLMs: Attacks, Defense, and Opportunities
Sunder Ali Khowaja, Parus Khuwaja, Kapal Dev, Hussam Al Hamadi, Engin Zeydan
Large Language Model Membership Inference Attack New Attack Biological Pathway Model Privacy

July 30, 2024

Adaptive Pre-training Data Detection for Large Language Models via Surprising Tokens
Anqi Zhang, Chaofeng Wu
Pre Training Membership Inference Attack K TOKEN Pre Training Corpus Opaque Machine Learning Pre Training Data Detection

July 26, 2024

July 24, 2024

July 22, 2024

Synthetic Image Learning: Preserving Performance and Preventing Membership Inference Attacks
Eugenio Lomurno, Matteo Matteucci
Synthetic Data System Performance Membership Inference Attack Generative Question Synthetic Training Image Data Recycling Latent Distillation

Membership Inference Attack

Papers

Membership Inference Attacks Cannot Prove that a Model Was Trained On Your Data

Survey of Security and Data Attacks on Machine Unlearning In Financial and E-Commerce

Subject Data Auditing via Source Inference Attack in Cross-Silo Federated Learning

Order of Magnitude Speedups for LLM Membership Inference

Unleashing Worms and Extracting Data: Escalating the Outcome of Attacks against RAG-based Inference in Scale and Severity Using Jailbreaking

Generated Data with Fake Privacy: Hidden Dangers of Fine-tuning Large Language Models on Generated Data

Context-Aware Membership Inference Attacks against Pre-trained Large Language Models

Understanding Data Importance in Machine Learning Attacks: Does Valuable Data Pose Greater Harm?

Membership Inference Attacks Against In-Context Learning

MIA-Tuner: Adapting Large Language Models as Pre-training Text Detector

Membership Inference Attack Against Masked Image Modeling

Range Membership Inference Attacks

Privacy-Preserving Split Learning with Vision Transformers using Patch-Wise Random and Noisy CutMix

Pathway to Secure and Trustworthy 6G for LLMs: Attacks, Defense, and Opportunities

Adaptive Pre-training Data Detection for Large Language Models via Surprising Tokens

Accuracy-Privacy Trade-off in the Mitigation of Membership Inference Attack in Federated Learning

Granularity is crucial when applying differential privacy to text: An investigation for neural machine translation

Explaining the Model, Protecting Your Data: Revealing and Mitigating the Data Privacy Risks of Post-Hoc Model Explanations via Membership Inference

Can Watermarking Large Language Models Prevent Copyrighted Text Generation and Hide Training Data?

Synthetic Image Learning: Preserving Performance and Preventing Membership Inference Attacks