colorado-mesa-university-cybersecurity / deeplearning-ids Goto Github PK

how to correct the errror

Hi, could you provide me a CSE-CIC-IDS2018 dataset? I can not download it on AWS.

Is loss correct for this row?

Clay, since you're using % attacks correctly classified, etc., can you update the values up to 2 decimal point accuracy? Tables usually need more accurate results.

Traceback (most recent call last):
File "fastai-expriments.py", line 94, in
experimentIndividual(sys.argv[1])
File "fastai-expriments.py", line 54, in experimentIndividual
procs = [FillMissing, Categorify]
NameError: name 'FillMissing' is not defined

Lucas, what's the progress in this section? See the IDS Paper if you're not aware of it.

Lucas, how's the progress with Keras-Theano? Can you update the results if you're done?

create confusion matrices for multi-class and binary-class experiments (from one experiment is good)

Make sure to record GPU times for each experiment

Clay, can you calculate * values for the table: Attack Sample Distribution and Classification Accuracy, ASAP?

| 02-15 | n/a | n/a | n/a
Why n/a for this row, Clay?

Hi,

Im new to AI/ML/Deeplearning. I am following this repository and was able to perform on training the model I believe. I was wondering If how can I test these models with a real-world network traffic or based from the downloaded sample data. Something like it would tell me that a particular traffic will be flag as brute force or something. Kind of like this one https://github.com/rambasnet/DeepLearning-IDS#attack-sample-distribution-and-detection-accuracy

Can you email me those to complete the paper? Ray, you can use your CU Boulder affiliation.
I need them sooner to complete the paper. Also, Ray are you keeping an eye on the CPU experiment running in CRL server?