Joint Mutual Information Maximimisation (JMIM) CLI Application

A bit of context

This repository follows after my final year project. The aim of the project was to reduce the memory consumption of a LightGBM model used for malware detection. To accomplish this, I used the Joint Mutual Information Maximisation (JMIM) algorithm to select the $K$ most 'important' variables to train the model on. Given there was no public JMIM pre-built package at the time, the algorithm was written in Python from scratch.

The Problem

My original implementation has the following issues:

1. Not language agnostic i.e., only suitable for Python programs.
2. Adopts a slow, naive approach to handle discrete and continuous variables.

The Solution

The purpose of this repository is to use JMIM to create a tool which selects the $K$ most 'important' variables via CLI given a CSV file of discrete/continuous random variables.

A dash of mathematics

For those interested to know how the algorithm above was derived go to section 3.4.1 of my final year project paper.

Installation

For Windows users:

1. Clone the repository: git clone https://github.com/DeraUchenwoke/jmim.git
2. In PowerShell terminal run: cd scripts followed by .\setup.ps1.

Usage

Examples

Style guide & practices

Google style guide. Powershell practice and style.

The tool was written in line with CPP core guidelines.