Network Intrusion Detection in an Adversarial Setting

Broad breakdown of the work involved in the project -

• Understanding Adversarial Machine Learning
• Understanding and using the Cleverhans Python module
• Trying to fool Machine Learning based classifiers for NIDS by making them classify malicious network traffic as benign.

---

Requirements

• Python 3.x +
• Cleverhans
• Keras
• Tensorflow
• numpy
• pandas
• scikit_learn
• matplotlib

The exact versions of the modules can be found in the requirements.txt file.

Steps to run

Since the code is still in the developmental phase, the current best version of the code is in the test/ folder. To run enter the following commands on the terminal

$ sudo pip3 install -r requirements.txt
$ cd test
$ python3 for_graph.py

# If you want to see the training accuracy with epochs, run
$ python3 draw_graph.py