Power iteration for convolution

deel-ai / deel-lip

Build and train Lipschitz constrained networks: TensorFlow implementation of k-Lipschitz layers

https://deel-ai.github.io/deel-lip/

MIT License

89 stars 10 forks source link

Power iteration for convolution #70

Closed cofri closed 1 year ago

cofri commented 1 year ago

The power iteration algorithm is introduced to compute the maximum singular value of a convolution. The function spectral_normalization_conv() returns a normalized convolution kernel.