Chain rule proof

22.3. Single Variable Calculus — Dive into Deep Learning 1.0.0-beta0 documentation Can someone please explain the third line? Like how it was derived
image

It uses the property that:

eq

The tricky thing is that f in this case would be g, x would be h(x) and epsilon would actually be epsilon * dh/dx(x)

1 Like

Thanks Brian! It is very much clear now! :smiley:

1 Like