Towards the Resistance of Neural Network Watermarking to Fine-tuning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913816203231232 |
|---|---|
| author | Tang, Ling Chen, Yuefeng Xue, Hui Zhang, Quanshi |
| author_facet | Tang, Ling Chen, Yuefeng Xue, Hui Zhang, Quanshi |
| contents | This paper proves a new watermarking method to embed the ownership information into a deep neural network (DNN), which is robust to fine-tuning. Specifically, we prove that when the input feature of a convolutional layer only contains low-frequency components, specific frequency components of the convolutional filter will not be changed by gradient descent during the fine-tuning process, where we propose a revised Fourier transform to extract frequency components from the convolutional filter. Additionally, we also prove that these frequency components are equivariant to weight scaling and weight permutations. In this way, we design a watermark module to encode the watermark information to specific frequency components in a convolutional filter. Preliminary experiments demonstrate the effectiveness of our method. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_01007 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Towards the Resistance of Neural Network Watermarking to Fine-tuning Tang, Ling Chen, Yuefeng Xue, Hui Zhang, Quanshi Machine Learning Artificial Intelligence Computation and Language Computer Vision and Pattern Recognition This paper proves a new watermarking method to embed the ownership information into a deep neural network (DNN), which is robust to fine-tuning. Specifically, we prove that when the input feature of a convolutional layer only contains low-frequency components, specific frequency components of the convolutional filter will not be changed by gradient descent during the fine-tuning process, where we propose a revised Fourier transform to extract frequency components from the convolutional filter. Additionally, we also prove that these frequency components are equivariant to weight scaling and weight permutations. In this way, we design a watermark module to encode the watermark information to specific frequency components in a convolutional filter. Preliminary experiments demonstrate the effectiveness of our method. |
| title | Towards the Resistance of Neural Network Watermarking to Fine-tuning |
| topic | Machine Learning Artificial Intelligence Computation and Language Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2505.01007 |