Safe and Robust Watermark Injection with a Single OoD Image

TL;DR

This paper introduces a novel watermarking method for deep neural networks that uses a single out-of-distribution image as a secret key, ensuring data privacy, robustness against attacks, and efficiency without requiring training data.

Contribution

It presents a safe, robust, and data-agnostic watermarking technique leveraging a single OoD image and parameter perturbation for IP protection in neural networks.

Findings

Effective watermark verification using a single OoD image.

Robustness against fine-tuning, pruning, and model extraction attacks.

Time- and sample-efficient without training data.

Abstract

Training a high-performance deep neural network requires large amounts of data and computational resources. Protecting the intellectual property (IP) and commercial ownership of a deep model is challenging yet increasingly crucial. A major stream of watermarking strategies implants verifiable backdoor triggers by poisoning training samples, but these are often unrealistic due to data privacy and safety concerns and are vulnerable to minor model changes such as fine-tuning. To overcome these challenges, we propose a safe and robust backdoor-based watermark injection technique that leverages the diverse knowledge from a single out-of-distribution (OoD) image, which serves as a secret key for IP verification. The independence of training data makes it agnostic to third-party promises of IP security. We induce robustness via random perturbation of model parameters during watermark injection to defend against common watermark removal attacks, including fine-tuning, pruning, and model extraction. Our experimental results demonstrate that the proposed watermarking approach is not only time- and sample-efficient without training data, but also robust against the watermark removal attacks above.