Variational Recurrent Neural Network

Variational Recurrent Neural Network implemented by Tensorflow

Last updated: Mar. 1st, 2018. Author: Tatsuro Yamada <ymt2.casino@gmail.com>

Requirements

• Python 2.7 (NO supports for 3.4 nor 3.5)
• Tensorflow 1.4
• NumPy 1.11

Implementation

• The model used in Denton and Fergus, "Stochastic Video Generation with a Learned Prior" blog

Example

$ cd train
$ python ../src/learn.py
  (It may take several minutes to download moving MNIST dataset for the first time)
$ python ../src/generate.py

Following must be taken into consideration

• Reconstruction error (mse? cross entropy?)
• Beta (the balance between reconstruction and regularization)
• Encoder and decoder (end-to-end? importing pretrained model?)