the lecture, we learned that an efficient implementation of backpropagation leverages the reuse

of previously computed partial derivatives. Which quantity could you reuse $($we store the common

terms in a single matrix$/$vector$)$ amongst the three partial derivatives calculated above to minimize

duplicate computation?

Write your answer in terms of

Uo$,\{$w$1,...,$wK $\}=[$uo$,$uw$1,...,$uwK

$],$ a matrix with the outside vectors stacked as columns,

and $1,$ a $($K $+1)\backslash$times $1$ vector of $1$s$\mathrm{.}8$ Additional terms and functions $($other than Uo$,\{$w$1,...,$wK $\}$ and

$1)$ can be used in your solution.