Too Long; Didn't Read
Google Deepmind’s paper DRAW: A Recurrent Neural Network For Image Generation. The code is based on the work of Eric Jang, who in his original code was able to achieve the implementation in only 158 lines of Python code. DRAW networks combine a novel spatial attention mechanism that mimics the foveation of the human eye, with a sequential variational auto-encoding framework that allows for the iterative construction of complex images. The system substantially improves on the state of the art for generative models on MNIST.