Very good issue

[yue123161]

It is a good idea to use self supervised learning for intrusion detection But how do you preprocess CIC-IDS 2017 dataset into pickle files, could you share the data preprocessing script .

[Johnsen92]

Hey,

thank you for the encouragement. I used the data preprocessing from this project https://github.com/CN-TU/adversarial-recurrent-ids/

The preprocessing is done in https://github.com/CN-TU/adversarial-recurrent-ids/blob/master/parse.py

Best, Jonas

[yue123161]

Hey, Jonas Thanks for your replay, I have download flow files from CN-TU but still there lacks some json files like args.json and _categories_mapping.json How could i setting those json files to run this code

Regards Darwin

[Johnsen92]

Hey Darwin,

I pushed my latest changes, did some cleanup and made the project more userfriendly. If you pull now you have all the necessary folders to run the code. You will have to unpack and copy flows.pickle and flows_categories_mapping.json from the root folder of https://github.com/CN-TU/adversarial-recurrent-ids/ into the ./data folder. Then you can just run "make lstm" or "make transformer" to run with default parameters. Note: The project is hard coded to run the training on a single GPU. If your setup is different, you will have to make some changes.

Best, Jonas

[yue123161]

Hey, Jonas Thanks for your comment, I have already running the code following your instruction. Hope can discuss you further about the principle behind the code. Regards Darwin

[Johnsen92]

Hey Darwin,

glad to hear that. The project is still in development and I will push updates frequently. If you have any suggestions or pointers don't hesitate to send them to me under e1226597@student.tuwien.ac.at.

Best, Jonas