Auditing Data Provenance in LLM Fine-tuning via Intrinsic Distributional Fingerprints
Open Access
- 14 November 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 6431-6445
Abstract
No abstract availableFunding Information
- NSFC (62272222, 62272215)
- Jiangsu Province Outstanding Youth Fund project (BK20230080)
- Nanjing University-China Mobile Communica- tions Group Co.,Ltd. Joint Institute (NJ20250025)
This publication has 14 references indexed in Scilit:
- Position: Membership Inference Attacks Cannot Prove That a Model was Trained on Your Data2025
- SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)2025
- zkLLM: Zero Knowledge Proofs for Large Language Models2024
- How to Protect Copyright Data in Optimization of Large Language Models?Proceedings of the AAAI Conference on Artificial Intelligence, 2024
- Analyzing Leakage of Personally Identifiable Information in Language Models2023
- Distribution Inference Risks: Identifying and Mitigating Sources of Leakage2023
- Text Embeddings Reveal (Almost) As Much As Text2023
- Self-Instruct: Aligning Language Models with Self-Generated Instructions2023
- Poisson mixturesNatural Language Engineering, 1995
- Use of the Correlation Coefficient with Normal Probability PlotsThe American Statistician, 1985