= Train a neural network N(x) in Tensorflow that can approximate f(x) = sin(x) + cos(x) with an average 2-norm of less than 0.02 using the objective function tf.reduce_mean(tf.linalg.norm (N(x)-f(x))) where N(x) is the output of your neural network and f(x) is the ground truth function. Generate at least 20 datapoints to train your neural network and evaluate with at least 20 datapoints as your test dataset. How many layers and how wide were the layers in your final solution? Use pyplot or a similar Python package to plot your neural network prediction N(x), colored blue, versus the ground truth labels of f(x), colored red