Algorithm 1: GTAD Training Algorithm

Input: Training Dataset X = {x1, …, xM}, The number of epochs R

Output: Trained GTAD

GTAD←initialize weight

epoch←1

repeat

for t = L to M do

$x{\prime }_t^i$$\mathrm{r}{x}_{t-l:t-1}^{\prime i}$← TAD (xt-L:t-1)

$Loss=\sqrt{x_t^i-x{\prime }_t^i}+\sqrt{{\displaystyle \sum }_{j=t-L}^{t-1}\left(x_j^i-x{\prime }_j^i\right)}$

GTAD←update weight using Loss

end for

epoch←epoch + 1

until epoch = R