Time series learning applied to intrusion detection in IT systems

Table of content

• Description
• Report
• Requirements
• Project
  • Network logs classification
  • Text prediction
  • Learning finite state machine
  • Cybersecurity automation
• Author

Description

This project is a 2nd year engeneering school internship project at the University Of Québec in Outaouais.
The purpose was to search and developp a program to code a neural network to learn time series in order to automate the sort of security alerts in a network.

Report

You can access and read my final report in the folder report.

Requirements

To install all required libraries, go to project root and run:

$ pip install -r requirements.txt

Project

Network logs classification

To execute the logs classification, go to src/Intrusion-Detection-NSL-KDD/ folder.
Then, use the command :

$ python3 attack_classification.py

Text prediction

To execute the text prediction, go to src/Word-Prediction folder.
Then, use this command to fit the model :

$ python3 text_precition.py

Then, use this command to test it :

$ python3 newt_word.py

You have to enter 4 words in order to predict the fifth one.

Learning finite state machine

To execute the text prediction, go to src/Finite-State-Machine folder.
Then, use this command to fit the model :

$ python3 neural_network.py

Then, use this command to test it :

$ python3 test_prediction.py

Cybersecurity automation

To execute the text prediction, go to src/Final-Project folder.
Then, use this command to fit the model :

$ python3 lstm.py

Authors

• MARAIS Lucas

MIT License - Copyright (c) 2022 Lucas MARAIS