https://github.com/kiannaquines/web-application-firewall-machine-learning

A simple web application firewall written using python with Machine Learning Algorithms
https://github.com/kiannaquines/web-application-firewall-machine-learning

Last synced: 2 months ago
JSON representation

A simple web application firewall written using python with Machine Learning Algorithms

• Host: GitHub
• URL: https://github.com/kiannaquines/web-application-firewall-machine-learning
• Owner: kiannaquines
• Created: 2025-01-26T06:36:29.000Z (4 months ago)
• Default Branch: main
• Last Pushed: 2025-02-03T13:56:06.000Z (4 months ago)
• Last Synced: 2025-02-03T14:32:40.024Z (4 months ago)
• Language: Jupyter Notebook
• Size: 23.2 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
## Web Application Firewall

This is my own version of implementing a web application firewall using machine learning algorithm.


![Image](https://github.com/user-attachments/assets/0f4e5ac3-c9bf-4524-ba1c-14a5cb99697f)

![Image](https://github.com/user-attachments/assets/cf4991c8-c930-4cf5-b198-533629cb175f)

## Model

Fitting the model in each algorithm takes 4 hours because of crappy specs of my laptop (CPU Based). It was so slow in fitting since the dataset that the model was trained has 100k+ records no GPU. 😅

The model was trained on both DecisionTreeClassifier & RandomForest, each algorithm got there own accuracy score.


![Image](https://github.com/user-attachments/assets/167fe288-6674-47d1-a71e-427b866446e2)

# Result

![Image](https://github.com/user-attachments/assets/1b9ec1fa-a955-4597-b81e-fd03b3857a89)

Out of 222 sql injection payload request 6 are missed to block the request.


![Image](https://github.com/user-attachments/assets/d9e6cd97-8a1b-47e0-894e-986dbc387872)

Out of 64 normal request 2 are missed to allow the request.


![Image](https://github.com/user-attachments/assets/9109e76b-0cee-44c6-b9ac-5c33d53e7e4e)

## Take Aways

1. Since some of the payloads are missed in classifiying whether the request is Malicious or Not we must include the missed payload in the training and testing phase.

2. We can use RNN or CNN for this web application firewall.

## Future Integration 

1. Integrate a Web Dashboard

2. Implement using Neural Network