Question

Why is the bptt chain rule different from the chain rule in maths?

Answer 1

The chain rule may be written in Leibniz's notation in the following way. If a variable zdepends on the variable y, which itself depends on the variable x, so that y and z are therefore dependent variables, then z, via the intermediate variable of y, depends on x as well. The chain rule then states,

{\displaystyle {\frac {dz}{dx}}={\frac {dz}{dy}}\cdot {\frac {dy}{dx}}.}