​ First, install PyTorch meeting your environment (recommmended 2.00):

pip3 install torch==2.0.0+cu111 torchvision==0.15.0+cu111 torchaudio==2.0.0+cu111 -f https://download.pytorch.org/whl/cu111/torch_stable.html

​ Then, use the following command to install the rest of the libraries:

pip3 install numpy pandas sklearn matplotlib scipy copy pathlib pyod

Please refer to the attached environment.yaml for our detailed experimental settings.

conda env create -f environment.yaml

​

• The MSL dataset is available in the data folder in the form of CSV files, and they are divided into a training dataset and a test dataset. ​
• main.py will automatically load dataset for training and testing ​

• The model evaluation metrics will be printed and logged in the form of a log. In the case of training, the training logs will also be recorded alongside. ​

• If you provide False to the last parameter of the parser of run.sh as -use_pretrained, the training and test will start without loading the pretrained model.

• Train and test the model(proactive_anomaly_detection) for the dataset MSL using GPU 0 (cpu for cpu) without pretrained model(False)

run.sh 0 MSL proactive_anomaly_detection False

• Pretrained model is in the pretrained folder

• If you provide True to the last parameter of the parser, main.py will load pretrained model and skip training ​

• test the model(proactive_anomaly_detection) for the dataset MSL using GPU 0 (cpu for cpu) with pretrained model(True)

run.sh 0 MSL proactive_anomaly_detection True