Collection of machine learning related notebooks to share.
Investigates the behavior of 2D FFTs on various test signals and images. It provides routines to preprocess and plot FFTs and inverse FFTs for complex signals as well as for real signals. Finally, it shows the impact on the reconstructed images with applied filters in the Fourier domain. Check out my article, Fourier CNNs with Kernel Sizes of 1024x1024 and Larger that has been published together with this notebook for more details about fourier convolutions.
In this Notebook, TensorFlow's tutorial on the DCGAN is adapted to be able to train the model on TPU. Check this guide for details on distributed training on TPU.
In this notebook the effect of the temperature value of the softmax transformation is investigated.
since 16.04.2022