About anomaly detection。

[ANopony]

hello，你好。
I am a little confused about the anomaly detection part.
It seems that the code read the abnormal interval‘s border from filename, which are partly abnormal, and cut the interval into many parts( length = patch_len), then caculate MSE between ground_truth and prediction fo each part. Finaly the part that has max MSE is right the abnormal data. Am I get the point?
If I want to use my custom dataset to test, I have to give the a interval that is partly abnormal first, right?
Thank you for your patient answer to such a stupid question.

[WenWeiTHU]

Thanks for your interest. I think you are right. Please refer to the paper for the detailed usage pipeline of the UCR anomaly datasets.