PNN

An Implementation Of Progressive Neural Networks for Intrusion Detection Systems

This repository hosts a basic implementation of Progressive Neural Networks(PNN) as stated in the original paper (https://arxiv.org/pdf/1606.04671.pdf).
A great part of this project is based on the efforts made by Maxwell J Jacobson in creating Doric using PyTorch.
This implementation will be used in a research project conducted in the Information, Network, and Learning Lab (INL) of Sharif University of Technology. After the research paper is published, further experiments and results will be provided.
Going through the code, you might notice that a data.py file is missing. This file contains a controller class which function is to generate train, validation, and test data (with respect to the configured batch size) from the vectorized flows obtained from the CIC-IDS2017 dataset.
Since the data used for the experiments is exclusive to the INL, this file can't be shared before the paper gets published. Nevertheless, implementing a controller isn't tricky, so you can write your own code for the controller based on your data and replace it in the code.