Depends on #83
The winning lottery ticket L is a sub-graph of the trained model M. If L[i] (resp. M[i]) is layer i of L (resp. M), it would be interesting to see if the ratio R[i]=len(set(L[i]).intersection(M[i])) / len(M[i]) is correlated to the robustness of layer i.
Maybe first check what they did in https://www.cs.montana.edu/sheppard/pubs/ijcnn-2019d.pdf
Depends on #83 The winning lottery ticket
L
is a sub-graph of the trained modelM
. IfL[i]
(resp.M[i]
) is layeri
ofL
(resp.M
), it would be interesting to see if the ratioR[i]=len(set(L[i]).intersection(M[i])) / len(M[i])
is correlated to the robustness of layeri
. Maybe first check what they did in https://www.cs.montana.edu/sheppard/pubs/ijcnn-2019d.pdf