NN Model

The NN model should use previous functions in the right order.

def nn_model(X, Y, n_h, num_iterations = 10000, print_cost=False):
     """
     Arguments:
     X -- dataset of shape (2, number of examples)
     Y -- labels of shape (1, number of examples)
     n_h -- size of the hidden layer
     num_iterations -- Number of iterations in gradient descent loop
     print_cost -- if True, print the cost every 1000 iterations
     Returns:
     parameters -- parameters learnt by the model. They can then be used to predict
     """

     return parameters

Parameters initialized

Initialize parameters, then retrieve W1, b1, W2, b2. Inputs: “n_x, n_h, n_y”

parameters = initialize_parameters(n_x, n_h, n_y)

Cost for every 1000 iterations

if print_cost and i % 1000 == 0:
    print ("Cost after iteration %i: %f" %(i, cost))

Expected output

  • cost after iteration 0 - 0.692739

Then the value of W1, b1, W2, b2