High-Robustness, Low-Transferability Fingerprinting of Neural Networks
This work addresses the need for practical and reliable fingerprinting methods in machine learning to reduce false positives, though it appears incremental by building on existing fingerprinting concepts.
The paper tackled the problem of fingerprinting deep neural networks by proposing Characteristic Examples that achieve high robustness against model pruning and low transferability to unrelated models, resulting in a new metric called Uniqueness Score to measure this trade-off.
This paper proposes Characteristic Examples for effectively fingerprinting deep neural networks, featuring high-robustness to the base model against model pruning as well as low-transferability to unassociated models. This is the first work taking both robustness and transferability into consideration for generating realistic fingerprints, whereas current methods lack practical assumptions and may incur large false positive rates. To achieve better trade-off between robustness and transferability, we propose three kinds of characteristic examples: vanilla C-examples, RC-examples, and LTRC-example, to derive fingerprints from the original base model. To fairly characterize the trade-off between robustness and transferability, we propose Uniqueness Score, a comprehensive metric that measures the difference between robustness and transferability, which also serves as an indicator to the false alarm problem.