Watermarking Kolmogorov-Arnold Networks for Emerging Networked Applications via Activation Perturbation

TL;DR

This paper introduces DCT-AW, a novel watermarking technique for Kolmogorov-Arnold Networks that embeds watermarks via activation perturbation, ensuring robustness and task independence in protecting model intellectual property.

Contribution

The paper presents the first watermarking method specifically designed for KAN, leveraging activation perturbation with DCT to enhance robustness and compatibility.

Findings

DCT-AW maintains high model performance.

It is robust against pruning and retraining attacks.

It outperforms existing watermarking methods.

Abstract

With the increasing importance of protecting intellectual property in machine learning, watermarking techniques have gained significant attention. As advanced models are increasingly deployed in domains such as social network analysis, the need for robust model protection becomes even more critical. While existing watermarking methods have demonstrated effectiveness for conventional deep neural networks, they often fail to adapt to the novel architecture, Kolmogorov-Arnold Networks (KAN), which feature learnable activation functions. KAN holds strong potential for modeling complex relationships in network-structured data. However, their unique design also introduces new challenges for watermarking. Therefore, we propose a novel watermarking method, Discrete Cosine Transform-based Activation Watermarking (DCT-AW), tailored for KAN. Leveraging the learnable activation functions of KAN, our method embeds watermarks by perturbing activation outputs using discrete cosine transform, ensuring compatibility with diverse tasks and achieving task independence. Experimental results demonstrate that DCT-AW has a small impact on model performance and provides superior robustness against various watermark removal attacks, including fine-tuning, pruning, and retraining after pruning.