uozbek / deep-learning-nids

A Deep Learning approach toward creating a NIDS using python

Network intrusion detection system using deep learning

Deep Learning Models for NIDS using NSL-KDD and ICIDS2017 datasets.

Requirements

• python 3.7
• pipenv module

Installation

Use pipenv install to install dependencies and pipenv shell to run the virtual environment.

Models

Each model has its own file with this format {model_name}.py.

• LSTM model
• GRU model
• RNN model
• DNN model
• Classic machine learning models (Naive Bayes, Ada Boost and more)

Usage

• Create a data directory at the root of the project if not exists.
• Put NSL-KDD dataset into data/nsl directory
• Put CICIDS2017 dataset into data/cicids/ directory
• Depending on your choices, these directories should be created into data directory: mul-nsl, mul-cicids, bin-nsl and bin-cicids.
• Run for each model using python run.py

Notes on running

• CICIDS2017 is a pretty large dataset and processing it is time and memory consuming so for test purposes you can use the NSL dataset.