PoLO: Proof-of-Learning and Proof-of-Ownership at Once with Chained Watermarking
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915953528274944 |
|---|---|
| author | Deng, Haiyu Jiang, Yanna Yu, Guangsheng Wang, Qin Wang, Xu Ma, Baihe Ni, Wei Liu, Ren Ping |
| author_facet | Deng, Haiyu Jiang, Yanna Yu, Guangsheng Wang, Qin Wang, Xu Ma, Baihe Ni, Wei Liu, Ren Ping |
| contents | Our evaluation shows that PoLO achieves \textbf{99\%} watermark detection accuracy for ownership verification, while preserving data privacy and cutting verification costs to just \textbf{1.5--10\%} of traditional methods. Forging PoLO demands \textbf{1.1--4$\times$} more resources than honest proof generation, with the original proof retaining over \textbf{90\%} detection accuracy even after attacks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_12296 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | PoLO: Proof-of-Learning and Proof-of-Ownership at Once with Chained Watermarking Deng, Haiyu Jiang, Yanna Yu, Guangsheng Wang, Qin Wang, Xu Ma, Baihe Ni, Wei Liu, Ren Ping Cryptography and Security Artificial Intelligence Machine Learning Our evaluation shows that PoLO achieves \textbf{99\%} watermark detection accuracy for ownership verification, while preserving data privacy and cutting verification costs to just \textbf{1.5--10\%} of traditional methods. Forging PoLO demands \textbf{1.1--4$\times$} more resources than honest proof generation, with the original proof retaining over \textbf{90\%} detection accuracy even after attacks. |
| title | PoLO: Proof-of-Learning and Proof-of-Ownership at Once with Chained Watermarking |
| topic | Cryptography and Security Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2505.12296 |