Covidulator

December ’20 || Unreal C++ Python Machine Learning

I integrated a machine-learning model into a VR Experience to learn how to use TCP/IP to communicate between different apps.

Motivation

How does my Alexa respond to me when it clearly isn’t running a Python program natively?

When taking a class on Machine Learning and Interaction, I never got a reasonable answer on how ML is used in the real world. How does my Alexa respond to me when it clearly isn’t running a Python program natively?

To answer this question, I created THE COVIDULATOR for my final project for the class. It’s a VR experience that uses Unreal to communicate with a machine-learning model in a Python program. Here’s the source code.

The Covidulator calculator uses Machine Learning to “compute” solutions

Execution

I created an ML model using the K-Nearest Neighbors (KNN) algorithm to predict patient survival from COVID-19. I chose to use KNN as it is simple to implement and can make predictions about the grouping of an individual data point by looking at relevant data to it.

I put the model into a TCP/IP server and created a way to interface it through parameters given by the TCP/IP client.

Then, I created a TCP/IP client in Unreal Engine using C++. I then created a keyboard that would input parameters into the model.