Consider a neural network with $3$ inputs, one hidden layer with $5$ units, and an output layer with $2$ units.

$$ A As a fully connected network this network should have $25$ connections, hence $25$ weights and $7$ bias terms.

$$ B The parameters of basic neural network can be estimated in closed form by setting the gradient of the cost function to zero.

$$ If trained this network could be used for binary classification.

$$ D The choice of activation functions does not matter in this case.