Deep learning vs shallow learning

Could you please help me to understand
I did non grasp why shallower NN requires exponentially more units in the case of xor operation. As for me, xor operation for n values requires n operations, not 2^n.
Will be grateful for the answer.

My best regards, Vasyl.

Hi,
can you please post this, in the course → week → video you are referring to?

I believe this thread is a duplicate.

Yes, this question was asked and answered on this other parallel thread.

Also note that other thread is correctly “tagged” with week 4, instead of week 1.