paper-with-me

Papers

Detecting Data Contamination in Large Language Models

2026-04-21 · Juliusz Janicki, Savvas Chamezopoulos, Evangelos Kanoulas, Georgios Tsatsaronis arxiv

Large Language Models (LLMs) utilize large amounts of data for their training, some of which may come from copyrighted sources. Membership Inference Attacks (MIA) aim to detect those documents and whether they have been included in the training corpora of the LLMs. The black-box MIAs require a significant amount of data manipulation; therefore, their comparison is often challenging. We study state-of-the-art (SOTA) MIAs under the black-box assumptions and compare them to each other using a unified set of datasets to determine if any of them can reliably detect membership under SOTA LLMs. In addition, a new method, called the Familiarity Ranking, was developed to showcase a possible approach to black-box MIAs, thereby giving LLMs more freedom in their expression to understand their reasoning better. The results indicate that none of the methods are capable of reliably detecting membership in LLMs, as shown by an AUC-ROC of approximately 0.5 for all methods across several LLMs. The higher TPR and FPR for more advanced LLMs indicate higher reasoning and generalizing capabilities, showcasing the difficulty of detecting membership in LLMs using black-box MIAs.

📄 PDF Abstract BibTeX arXiv:2604.19561

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

DICE: Detecting In-distribution Contamination in LLM's Fine-tuning Phase for Math Reasoning

2024-06-06 · Shangqing Tu, Kejian Zhu, Yushi Bai, Zijun Yao 외

The advancement of large language models (LLMs) relies on evaluation using public benchmarks, but data contamination can lead to overestimated performance. Previous researches focus on detecting contamination by determin…

Math

A Taxonomy for Data Contamination in Large Language Models

2024-07-11 · Medha Palavalli, Amanda Bertsch, Matthew R. Gormley

Large language models pretrained on extensive web corpora demonstrate remarkable performance across a wide range of downstream tasks. However, a growing concern is data contamination, where evaluation datasets may be con…

Question Answering

Detecting Benchmark Contamination Through Watermarking

2025-02-24 · Tom Sander, Pierre Fernandez, Saeed Mahloujifar, Alain Durmus 외

Benchmark contamination poses a significant challenge to the reliability of Large Language Models (LLMs) evaluations, as it is difficult to assert whether a model has been trained on a test set. We introduce a solution t…

ARCMMLU

Both Text and Images Leaked! A Systematic Analysis of Multimodal LLM Data Contamination

2024-11-06 · Dingjie Song, Sicheng Lai, Shunian Chen, Lichao Sun 외

The rapid progression of multimodal large language models (MLLMs) has demonstrated superior performance on various multimodal benchmarks. However, the issue of data contamination during training creates challenges in per…

Towards Data Contamination Detection for Modern Large Language Models: Limitations, Inconsistencies, and Oracle Challenges

2024-09-16 · Vinay Samuel, Yue Zhou, Henry Peng Zou

As large language models achieve increasingly impressive results, questions arise about whether such performance is from generalizability or mere data memorization. Thus, numerous data contamination detection methods hav…

Memorization