Start building the keras implementation of deepcover testing

README.md

@@ -4,10 +4,9 @@ Uncover Bugs in Deep Learning
 DeepCover is a coverage-based bug finder for deep learning applications.
 It is able to provide metrics for evaluating the robustness of a Deep Neural Network (DNN).
 The basic idea of DeepCover is easy. For any neuron in the DNN, if its state changes,
-there must be a cause for it: DeepCover aims to COVER these changes and causes.
+there must be a cause for it: DeepCover aims to COVER a change and its immediate causes.
 More details can be found in the paper [Testing Deep Neural Networks](https://arxiv.org/abs/1803.04792).
 
-## To run the tool, the following software or Python package is required:
-  theano, numpy, matplotlib, IBM CPLEX
+## To run the tool:
+  These routine machine learning Python packages are needed and IBM CPLEX is used as the Linear Programming solver
 
-Currently, DeepCover is under development to support TensorFlow.

src/deepcover_keras.py

@@ -0,0 +1,37 @@
+
+import argparse
+import sys
+
+import keras
+from keras.models import *
+from keras.layers import * 
+from keras import *
+
+
+def main():
+  parser=argparse.ArgumentParser(
+          description='DeepCover: Uncover Bugs in Deep Learning' )
+
+  parser.add_argument('model', action='store', nargs='+', help='The input neural network model (.h5)')
+
+  parser.add_argument(
+          '--cover', metavar='ss', action='store', help='The covering method: ss, sv, ds, dv', default='ss')
+
+  args=parser.parse_args()
+
+  model = load_model(args.model[0])
+
+  if not (args.cover in ['ss', 'sv', 'ds', 'dv']):
+    print ('Covering method cannot be recognized: ' + args.cover)
+    sys.exit(0)
+
+  print ('\n== WARNING == \n')
+  print (
+    'The input model:       ' + args.model[0] + '\n' +
+    'The covering method:   ' + args.cover  + '\n'
+    )
+  print ('This keras compatible implementation of DeepCover testing is currently under deverlopment...\n')
+  print ('=============\n')
+
+if __name__=="__main__":
+  main()