imperial-qore
About SWaT dataset test time is too long, can not produce results, what should I do
It seems like there is some issue with your system or pytorch installation as 3000+ seconds for training is too high.
When my cpu is R7 5800 one epoch time is 400+ seconds , because I can't use gpu, how should I solve the problem that I can't test on SWaT dataset?
I seem to have found the problem here in pot_eval.
It keeps looping here, how do I fix it？
I compared your code with the POT in OmniAnomaly and found that you have an extra parameter of lm, what does this parameter do?
I made the mistake of being lms getting smaller and smaller and can't get out in the loop
result, pred = pot_eval(lt, l, ls); preds.append(pred)
I found that in this line, when i is at a certain value, the values of lt and l are all 0, resulting in an infinite loop.
When I remove this part of the code, the result I get is {'FN': 5178, 'FP': 253109, 'Hit@100%': 1.0, 'Hit@150%': 1.0, 'NDCG@100%': 1.0, 'NDCG@150%': 1.0, 'ROC/AUC': 0.6324510394527012, 'TN': 142189, 'TP': 49443, 'f1': 0.2768548424500904, 'precision': 0.1634198418730195, 'recall': 0.9052012960390323, 'threshold': 0.5505070690110198}
安装似乎存在一些问题，因为 3000+ 秒的训练
Hello, I have encountered a similar problem. How do you solve this problem?
