readme links

README.md

@@ -1,6 +1,6 @@
 # Generative Adversarial Networks for generating Extended-MNIST samples
 ### Training
-One fully connected layer and two convolution layers for both the generator and the discriminator were used. The neural network was trained for *approximately* 2k epochs, over the *extended-MNIST dataset*, which contains handwritten digits as well as letters of the English Alphabet. 
+One fully connected layer and two convolution layers for both the generator and the discriminator were used. The Generative Adversarial Network ([arXiv:1406.2661](https://arxiv.org/abs/1406.2661)) was trained for *approximately* 2k epochs, over the *extended-MNIST dataset*, which contains handwritten digits as well as letters of the English Alphabet. 
 
 The performance of the network seemed to be very sensitive to the hyper-parameters and initializations, especially the learning rate. Too large a learning rate caused the GAN to diverge from the minimum, whereas too small almost always led to Mode Collapse.
 
@@ -10,3 +10,6 @@ The use of xavier initializer and relu layer (non leaky) gave best results.
 <img src="https://github.com/OrionMonk/GAN_EMNIST/blob/master/train.gif" width="50%">
 
 *Fig. Samples generated during training*
+
+### Usage
+The network was trained on 23x19 sized EMNIST images. For usage on a slightly different dataset, the dimension of the input in main.py and the dimension of the hidden layers in lib/generator.py and lib/discriminator.py need to be changed.