True$/$False:

$L_1-$regularisation reduces small weights more than $L_2-$regularisation.

All Boolean functions with $N$ inputs can be represented by a perceptron with one hidden

layer.

Using $g(b)=b$ as activation function and setting all thresholds to zero in a multilayer

perceptron allows you to solve some linearly inseparable problems.

Two hidden layers are sufficient to approximate any real$-$valued function with $N$ inputs

and one output in terms of a perceptron.