So there's an approach proposed by Salmani Tehranipoor and Plusqellic in 2009.

And this motivational example shows the basic idea.

What we have here is we have a hierarchical two input and gate.

Assume that all the primary inputs are equally likely to be 0 or 1.

And the two input and gate will give one-quarter chance produce a 1 and

three-quarter of the chance produce a 0.

Because this will be 1 if and only if both signals will be 1.

And if we keep on proceeds this for the second level, and

the gate to produce the 1, it requires both input here to be 1?

So that is one-quarter times another quarter.

So one over 16 it will be a 1.

Similarly, we have a identical structure at the bottom half.

So now for these two input and the gate.

For this one to be a 1, it will require both signals here and here to be 1.

So this one would be considered a rare event, because the probability of

this happening is 1 over 16th times 1 over 16th, which is one over 256.

So now let's see how we can remove this rare event.

Make it much more frequent.