Mask-based Membership Inference Attacks for Retrieval-Augmented Generation
October 26, 2024 Β· Declared Dead Β· π The Web Conference
"No code URL or promise found in abstract"
Evidence collected by the PWNC Scanner
Authors
Mingrui Liu, Sixiao Zhang, Cheng Long
arXiv ID
2410.20142
Category
cs.CR: Cryptography & Security
Cross-listed
cs.AI,
cs.IR
Citations
20
Venue
The Web Conference
Last Checked
4 months ago
Abstract
Retrieval-Augmented Generation (RAG) has been an effective approach to mitigate hallucinations in large language models (LLMs) by incorporating up-to-date and domain-specific knowledge. Recently, there has been a trend of storing up-to-date or copyrighted data in RAG knowledge databases instead of using it for LLM training. This practice has raised concerns about Membership Inference Attacks (MIAs), which aim to detect if a specific target document is stored in the RAG system's knowledge database so as to protect the rights of data producers. While research has focused on enhancing the trustworthiness of RAG systems, existing MIAs for RAG systems remain largely insufficient. Previous work either relies solely on the RAG system's judgment or is easily influenced by other documents or the LLM's internal knowledge, which is unreliable and lacks explainability. To address these limitations, we propose a Mask-Based Membership Inference Attacks (MBA) framework. Our framework first employs a masking algorithm that effectively masks a certain number of words in the target document. The masked text is then used to prompt the RAG system, and the RAG system is required to predict the mask values. If the target document appears in the knowledge database, the masked text will retrieve the complete target document as context, allowing for accurate mask prediction. Finally, we adopt a simple yet effective threshold-based method to infer the membership of target document by analyzing the accuracy of mask prediction. Our mask-based approach is more document-specific, making the RAG system's generation less susceptible to distractions from other documents or the LLM's internal knowledge. Extensive experiments demonstrate the effectiveness of our approach compared to existing baseline models.
Community Contributions
Found the code? Know the venue? Think something is wrong? Let us know!
π Similar Papers
In the same crypt β Cryptography & Security
R.I.P.
π»
Ghosted
R.I.P.
π»
Ghosted
The Limitations of Deep Learning in Adversarial Settings
R.I.P.
π»
Ghosted
Distillation as a Defense to Adversarial Perturbations against Deep Neural Networks
R.I.P.
π»
Ghosted
Spectre Attacks: Exploiting Speculative Execution
R.I.P.
π»
Ghosted
How To Backdoor Federated Learning
R.I.P.
π»
Ghosted
Evasion Attacks against Machine Learning at Test Time
Died the same way β π» Ghosted
R.I.P.
π»
Ghosted
Federated Learning: Strategies for Improving Communication Efficiency
R.I.P.
π»
Ghosted
In-Datacenter Performance Analysis of a Tensor Processing Unit
R.I.P.
π»
Ghosted
Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning
R.I.P.
π»
Ghosted