wanjinchang / tensorflowdeepautoencoder Goto Github PK

#Deep Autoencoder with TensorFlow

A selection of first layer weight filters learned during the pretraining

##Introduction The purpose of this repo is to explore the functionality of Google's recently open-sourced "sofware library for numerical computation using data flow graphs", TensorFlow. We use the library to train a deep autoencoder on the MNIST digit data set. For background and a similar implementation using Theano see the tutorial at http://www.deeplearning.net/tutorial/SdA.html.

The main training code can be found in autoencoder.py along with the AutoEncoder class that creates and manages the Variables and Tensors used.

##Setup It is expected that Python2.7 is installed and your default python version. ###Ubuntu/Linux

$ git clone https://github.com/cmgreen210/TensorFlowDeepAutoencoder
$ cd TensorFlowDeepAutoencoder
$ sudo chmod +x setup_linux
$ sudo ./setup_linux  # If you want GPU version specify -g or --gpu
$ source venv/bin/activate 

###Mac OS X

$ git clone https://github.com/cmgreen210/TensorFlowDeepAutoencoder
$ cd TensorFlowDeepAutoencoder
$ sudo chmod +x setup_mac
$ sudo ./setup_mac
$ source venv/bin/activate 

##Run To run the default example execute the following command. NOTE: this will take a very long time if you are running on a CPU as opposed to a GPU

$ python code/run.py

Navigate to http://localhost:6006 to explore TensorBoard and view training progress.

View of TensorBoard's display of weight and bias parameter progress.