Algorithm 2: DNN regression.

/* Configure DNN regression                          */ 1 regressor = learn.DNNRegressor(feature_columns, hidden_units = [100, 200, 100], optimizer =  tf.train.ProximalAdagradOptimizer( learning_rate = 0.1, l1_regularization_strength = 0.001),  activation_fn = tf.nn.sigmoid)  /* Train measured data up to 4000 times                    */ 2 input_training_fn ← (awgn_snr, awgn_bler) 3 regressor.fit(input_fn = input_training_fn, steps = 4000)  /* Predict of BLERs for test SNRs                       */ 4 input_reff_fn ← snr range 5 predictions = list(regressor.predict_scores(input_fn = input_reff_fn)) 6 regressed_bler = np.asarray(predictions)