diff --git a/README.md b/README.md
index a1f2a1a..5343935 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 # DeepMerge
 
-Code repository for the paper "DeepMerge: Classifying High-redshift Merging Galaxies with Deep Neural Networks", A. Ćiprijanovića, G.F. Snyder, B. Nord, J.E.G. Peek, Astronomy & Computing, submitted
+Code repository for the [paper](https://doi.org/10.1016/j.ascom.2020.100390) "DeepMerge: Classifying High-redshift Merging Galaxies with Deep Neural Networks", A. Ćiprijanovića, G.F. Snyder, B. Nord, J.E.G. Peek, Astronomy & Computing, Volume 32, July 2020, 100390
 
 
 ### Abstract
@@ -20,11 +20,11 @@ DeepMerge has 3 convolutional layers (with 3 pooling layers) and 3 dense layers,
 Images used can be found at https://doi.org/10.17909/t9-vqk6-pc80. Pristine and noisy images used in the paper can be found in SB00_augmented.npy and SB25_augmented.npy, respectively (label files: SB00_augmented_y.npy and SB25_augmented_y.npy). Raw (large) and resized images are also available for those who would like to try their own image formating and augmentation. Images we use have 2 filters (they mimic those available onboard the Hubble Space Telescope). For use with more complex neural networks we also have 3-filter files available (SB00_augmented_3FILT.npy and SB25_augmented_3FILT.npy, and corresponding labels SB00_augmented_3FILT_y.npy and SB25_augmented_3FILT_y.npy)
 
 ### Training
-Training is performed with early stopping (with validation loss being monitored). To run training and plot training and classification diagnostics use DeepMerge.ipynb and DeepMerge.ipynb files. 
+Training is performed with early stopping (with validation loss being monitored). To run training and plot training and classification diagnostics use DeepMerge.ipynb and DeepMerge-noisy.ipynb files. 
 ![](images/training.png)
 
 ### Galaxy properties
-You can also use DeepMerge.ipynb and DeepMerge.ipynb files to plot information about galaxy morphology and physical properties, which is also available for our images - concentration, M20 and stellar mass (there are other parameters available to be extracted from the catalogue we use - illustris_morphs_rf.txt).
+You can also use DeepMerge.ipynb and DeepMerge-noisy.ipynb files to plot information about galaxy morphology and physical properties, which is also available for our images - concentration, M20 and stellar mass (there are other parameters available to be extracted from the catalogue we use - illustris_morphs_rf.txt).
 
 ### Gradient-weighted Class Activation Maps (Grad-CAMs)
 Grad-CAMs show which pixels of the image were the most important for the classification into a particular class. This is important as a sanity check when using neural networks, but it can also show how different architectures "look" at images differently when doing the same classification. It is also important if we want to track how additional noise or image resizing impacts decidion making of our neural network. For example, DeepMerge trained on pristine and noisy images looks at different regions of the same galaxy for the classification (see image below).