Gating layer in tensorflow

I have a single input x and I want to activate different neural networks f(x) according to its value. For example:

if x in [-1,-0.5], y = f1(x)
if x in (-0.5,0], y = f2(x)
if x in (0,0.5], y = f3(x)
if x in (0.5,1], y = f4(x)

How can I do it?

Additionally, if I would like to train also the gating (i.e., without specifying the intervals), which structure suits my needs best?