##################################################################
All right reserved. (C) MOU Lili


The package can be freely used for non-commercial purposes.
If you would like to use for researches, please cite:

Lili Mou, et al., Learning Real-Valued Program Representations for Deep Learning, XXX, 2014




##################################################################

This package contrains the following contents.

1. nn/

The is the infrastucture of feed-forward neural networks (FFNN), including forward propagation, and back propagtaion.

You can construct any FFNN you like based on it. What you should do is 

1) Specify the connections,
2) Specify the cost function,
3) Forward propagate and back propagate!

See train/ for the detailed usages.


2. train/

This is the network for program representation learning. Run 

$ python train.py



3. parameter/

This folder contains the learned representations and the codes for empirical study. Run

$ python testVector.py


4. data/

This folder contains the dataset. Do NOT open the directory with graphical interfaces.