Neural network Basics: Part One (training)

neural-net-basics-01-05.py

for i in range(10000):
    # Forward propagte our input to get a prediction
    l1 = sigmoid(np.dot(X, W0))
    l2 = sigmoid(np.dot(l1, W1))
    
    # Estimate how much we missed
    l2_error = (y - l2) ** 2
    l2_delta = l2_error * sigmoid(l2, prime=True)
    
    # Back propagate error in layer 2 into layer 1
    l1_error = np.dot(l2_delta, W1.T)
    l1_delta = l1_error * sigmoid(l1, prime=True)
    
    # Update weights
    W1 += np.dot(l1.T, l2_delta)
    W0 += np.dot(X.T, l1_delta)