Ch Nitish Chandra (120050071)

Output will be written to a file output.csv. It prints accuracy on the
entire training set (without cross validation)

Implementation Details:
Backpropagation with cross entropy error function is minimised using stochastic gradient descent. Following are the parameters:
- Network Architecture
  1 input layer, 1 hidden layer of 10 nodes, 1 output layer
- Training rate eta = 0.05
- Each input row is used 10 times for training

Feature Selection:
57th feature is a scaled version of 55th feature. To reduce the number of variables, I removed the 57th feature.