Hi all,
Fairly simple notation question: the equations listed for Ex. 2 are as follows:
z^{(i)} = Wavg^{(i)} + b
a^{(i)} = softmax(z^{(i)})
\mathcal{L}^{(i)} = - \sum_{k = 0}^{n_y - 1} Y_{oh,k}^{(i)} * log(a^{(i)}_k)
*Note: these are not solutions, they are equations provided in the prompt.
I managed to complete the exercise, finding that the z and loss functions both use the dot product rather than element-wise multiplication. The notation for them in the prompt is different, however, with the z value’s calculation not including a ’ * ’ and the Loss calculation including a ’ * '.
Is there a convention for when the dot product / element-wise multiplication / cross product / other multiplication should be used? It’s 100% ok if there isn’t a convention, I’m just trying to understand if there is a convention.
Thank you all!