Closed gehuangyi20 closed 6 years ago
In the original paper section 5.2.1, the paper said
Detector II and detector I(see Table 3) used the $L^2$ and $L^1$ norm to measure reconstruction error, respectively.
However, in the following implementation, you use L1 for detector II and L2 for detector I. https://github.com/Trevillie/MagNet/blob/b115cf8798dc0c304ca0da34b2bdf1485da6f54f/test_defense.py#L14 https://github.com/Trevillie/MagNet/blob/b115cf8798dc0c304ca0da34b2bdf1485da6f54f/test_defense.py#L15
I am asking is this a just typo? Can you confirm which one should be the correct one.
It seems the code in the repo gives the better result.
Hi @gehuangyi20
As far as I recall, that should be a typo in the paper. Thanks for the notice.
In the original paper section 5.2.1, the paper said
However, in the following implementation, you use L1 for detector II and L2 for detector I. https://github.com/Trevillie/MagNet/blob/b115cf8798dc0c304ca0da34b2bdf1485da6f54f/test_defense.py#L14 https://github.com/Trevillie/MagNet/blob/b115cf8798dc0c304ca0da34b2bdf1485da6f54f/test_defense.py#L15
I am asking is this a just typo? Can you confirm which one should be the correct one.
It seems the code in the repo gives the better result.