y = np.array([1] * batch_size + [0] * int(noise_size))