fixed formatting

dogs/README.md

@@ -1,4 +1,4 @@
-== Convolutional Neural Networks for Dog Breed Identification ==
+__Convolutional Neural Networks for Dog Breed Identification__
 
 Using convolutional neural networks, this project aims to be able to distinguish what breed a dog is given its photograph.  Using five different breeds with a total of approximately
 25,000 training images, the neural network classifier achieves a classification rate of approximately 90%. 
@@ -13,13 +13,13 @@ as well as Boxers. This yields a vector of "probabilities" for each image - sinc
 each image is converted into a string of 10\*2 = 20 numbers between zero and one. A small fully connected neural network is then trained on this data in order
 to give a final multi-class prediction. The classification rate on an independent test set is approximately 90%.
 
-Files:
-    - neuralnetworks.py - implementation of neural networks and data processing using Theano
-    - mnist/            - example use of neuralnetworks.py library to recognize MNIST digits (reaches 98% recognition rate)
-    - all\_dogs.py      - train a multi-class classifier for different breeds
-    - dogs.py           - trains pairwise comparison networks
-    - pairs.py          - run pairwise comparison networks on all input data
-    - pairtrain.py      - train a post-processing step to convert the output of the pairwise networks into a multi-class prediction
+__Files:__
+* neuralnetworks.py - implementation of neural networks and data processing using Theano
+* mnist/            - example use of neuralnetworks.py library to recognize MNIST digits (reaches 98% recognition rate)
+* all\_dogs.py      - train a multi-class classifier for different breeds
+* dogs.py           - trains pairwise comparison networks
+* pairs.py          - run pairwise comparison networks on all input data
+* pairtrain.py      - train a post-processing step to convert the output of the pairwise networks into a multi-class prediction
 
-Data:
-    - imgs.tar.gz       - dog photos sorted by breed. Each photo was resized to a 64x64 3-color image (losing aspect ratio). Data were obtained using Petfinder.com API.
+__Data:__
+* imgs.tar.gz       - dog photos sorted by breed. Each photo was resized to a 64x64 3-color image (losing aspect ratio). Data were obtained using Petfinder.com API.