Decide whether the following sentences are True or False.
If False, explain why.

1. A Perceptron can
solve non-linearly separable problems.

2. Since the sigmoid
function outputs values in the range [0, 1], we usually normalize the target
values to the range [0, 1] to solve the regression task.

3. We use the sigmoid
function in the output layer to handle nominal values, because the sigmoid
function has non-linearity.

4. Since it is not
recommended to use both the sigmoid function and the Cross Entropy (CE) Loss,
we mainly use MSE loss instead of CE loss in the 2-class classification task.

5. In a multi-class
classification task, we create virtual outputs through one-hot encoding.

6. If we use the MSE loss to solve the
multi-class classification task, the computation of the neural network is
impossible, so we use the Cross Entropy Loss.

7. We use the sigmoid
function in the output layer to solve the multi-label classification task.