# create FeefForward Neural Network
feedforward <- function(X, Theta1,Theta2) {
#Add a column of 1s to X (the first column), and it becomes 'a1'.
  a1 <- cbind(1, X)
#Multiply by Theta1 and we have 'z2'.
  z2 <- a1 %*% t(Theta1)
#Compute the sigmoid() of 'z2', then add a column of 1's, and it becomes 'a2'
  a2 <- sigmoid(z2)
  a2 <- cbind(1, a2)
#Multiply by Theta2 and you have 'z3'.
  z3 = a2 %*%  t(Theta2)
#Compute the sigmoid() of z3, then add a column of 1's, and it becomes 'a3'
  a3 = sigmoid(z3);
  h=a3
}