PoLO: Proof-of-Learning and Proof-of-Ownership at Once with Chained Watermarking

TL;DR

PoLO is a watermarking method that achieves high detection accuracy and privacy preservation, significantly reducing verification costs and resisting attacks with minimal resource overhead.

Contribution

This paper introduces PoLO, a novel chained watermarking scheme that combines proof-of-learning and proof-of-ownership efficiently and securely.

Findings

Achieves 99% watermark detection accuracy

Reduces verification costs to 1.5-10% of traditional methods

Maintains over 90% detection accuracy after attacks

Abstract

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.