TensorFlow Experiments

Experimental implementations of various things with TensorFlow. Here GANs and VAEs were tested on MNIST. RNNs were tested on some time-series datasets, e.g. UCR Time Series Classification Archive or the trivial positive/negative-slope dataset generated by generate_trivial_datasets.py.

TensorFlow 1.x

• Generative Adversarial Networks (GANs)
  • CycleGAN - see separate repo
  • GAN using tfgan - see Unconditional GAN/
• Variational Autoencoders (VAEs) - see VAE/
• Recurrent Neural Networks (RNNs)
  • LSTM, GRU, SimpleRNN, etc. comparison - see RNN/
  • VRNN - see VRNN/
  • VRADA - see separate repo

TensorFlow 2.x

• tf2-hello-world
• tf2-GAN
• self-ensembling