Cracking White-box DNN Watermarks via Invariant Neuron Transforms.
Xudong PanMi ZhangYifan YanYining WangMin YangPublished in: KDD (2023)
Keyphrases
- white box
- black box
- digital images
- source code
- neural network
- watermarking scheme
- test cases
- test data
- digital watermarking
- affine transformation
- copyright protection
- multiresolution
- neuron model
- spread spectrum
- watermarking technique
- geometric transformations
- moment invariants
- watermarking algorithm
- affine invariant
- feature vectors
- open source
- training process
- data sets
- software systems
- spike trains
- reinforced concrete
- feature extraction
- source code metrics