KNN for Detecting Adversarial Attacks on Computer Networks

National Action Council for Minorities in Engineering(NACME) Google Applied Machine Learning Intensive (AMLI) at the University of Arkansas

Developed by:

• Ron Merculief - University of Alaska Anchorage
• Abraham Mitchell - University of Arkansas
• Claudia Mercado - University of Arkansas

Description

1. This program uses data downloaded from https://csr.lanl.gov/data/cyber1/ orignating from the Los Alamos National Laboratory and is put into a pandas DataFrame.
2. The data is labelled from the redteam.txt file by matching the four columns provided to the main DataFrame.
3. This program then utilizes hardcode to gather unique catagories from each column of the main dataset and turns them into unique IDs and scaled to 0 to 1 and put into a dictionary
4. The scaled and labelled data is then split into 5% training data and 1% testing data and fed into K Nearest Neighbor classifier
5. After the model is trained the program predicts the testing data and compares them with a confusion matrix, accuracy, and f1 score
6. The redteam.txt file was used for exploratory data analysis afterward

Usage instructions

1. Fork this repo
2. Change directories into your project
3. On the command line, type pip3 install requirements.txt
4. ....