huankoh / cst-gl Goto Github PK

Hi, I used the WADI_14days data set when reproducing this code. According to the running instructions you gave, I got the following error after running the run.py file:

Traceback (most recent call last):
  File "run.py", line 315, in <module>
    pointwise, early = main(i)
  File "run.py", line 297, in main
    indicator, prediction = anomaly_detector.scorer(args.pca_compo)
  File "/mnt/anomaly_dd.py", line 48, in scorer
    test_norm = error_sw_normalizer(full_abs,self.window_length,self.batch_size,len(self.test_obs))
  File "/mnt/anomaly_dd.py", line 102, in error_sw_normalizer
    batch_norm_error = (batch_error_mat - median) / iqr
ValueError: operands could not be broadcast together with shapes (128,127) (0,127) 

I don't know what's going on, and I sincerely hope you can give effective guidance, thank you!

Hello author, it's a great honor to read your paper, but I was confused for a long time in the pre-processing process of SWaT data sets, as you said, I can use the method in the GDN paper to process the data, but after running the GDN/scripts/ process_sswat. All I got were train.csv, test.csv, and list.txt. As far as I know, these three files cannot be run directly into your CST-GL model because your model requires train.npz, test.npz, and val.npz. May I ask how I need to process it now, or can I share the Swat data processing code? Because the processing process of WADI is different from that of SWaT data, I hope to ask for your help. Thank you very much!

Hi,

Foremost, very interesting work! It looks really cool and congrats with the publication!

I would like to investigate your MTCL module a bit more, and I wondered if it is possible to run it separately somehow.
Or is it that this module is 'embedded' into the whole training procedure?

Kind regards,

Stefan

您好，根据您的提示，在运行SMD 数据集出现了以下错误。
我不知道如何处理这个错误，真诚希望您能够给予些指导，谢谢

Hi, thank you for sharing your excellent work!
After preprocessing the WADI dataset, I encountered the following error when running the run.py file.

I've identified the issue at this line: early = early_detection_evaluation(labels, indicator, args.delays). However, I'm unsure how to fix it and would sincerely appreciate your guidance. Thank you！

Hi! Can you please specify the source of dimensions B1 and B2 in the Figure .1 SMD dataset, particularly which machine they originate from? This information is crucial for our analysis. Thank you!

Hi, thanks for sharing your excellent work!
I followed instructions to pre-process the SMD data set first. Then I got the following error after running the run.py file (python run.py --data data/machine-1-1 --expid machine-1-1 --delays [0,1,5,10,20,30,60] --num_nodes 38 --subgraph_size 10 --pca_compo 8):

I don't know what's going on, and I sincerely hope you can give effective guidance, thank you!