I'm also investigating unsupervised anomaly detection as well, and would like to try this method as it reports good performance compared to previous methods.
I have implemented the model according to the paper, but some settings are not described, such as network architectures (layer types and numbers, kernel size, stride, max pooling..etc..), I think they may largely affect my results.
I will be happy refer to this repository if I can have the codes.
Many thanks.
I'm also investigating unsupervised anomaly detection as well, and would like to try this method as it reports good performance compared to previous methods.
I have implemented the model according to the paper, but some settings are not described, such as network architectures (layer types and numbers, kernel size, stride, max pooling..etc..), I think they may largely affect my results.
I will be happy refer to this repository if I can have the codes.
Many thanks.