Summary of Semantic Membership Inference Attack Against Large Language Models, by Hamid Mozaffari and Virendra J. Marathe
Semantic Membership Inference Attack against Large Language Models
by Hamid Mozaffari, Virendra J. Marathe
First submitted to arxiv on: 14 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces Semantic Membership Inference Attack (SMIA), a novel approach that enhances the performance of Membership Inference Attacks (MIAs) by leveraging semantic content of inputs and perturbations. SMIA trains a neural network to analyze the target model’s behavior on perturbed inputs, capturing variations in output probability distributions between members and non-members. The authors conduct comprehensive evaluations on Pythia and GPT-Neo models using the Wikipedia dataset, demonstrating that SMIA significantly outperforms existing MIAs. For instance, SMIA achieves an AUC-ROC of 67.39% on Pythia-12B, compared to 58.90% by the second-best attack. This research contributes to the development of novel membership inference attacks and their applications in various domains. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper introduces a new way to figure out if data was used to train a model or not. The method is called Semantic Membership Inference Attack (SMIA) and it’s better than other methods at doing this job. SMIA uses a special kind of neural network that looks at how the model behaves when given different inputs. The authors tested SMIA on two types of models using Wikipedia data and found that it did much better than the previous best method. This research can be used to make sure that data is being used fairly and securely. |
Keywords
» Artificial intelligence » Auc » Gpt » Inference » Neural network » Probability
