Machine learning anti-cheating algorithm and a test against computer vision aimbot

Abstract

In this article, we introduced a recurrent neural network (RNN) serves as a cheating detection method in CS: GO against traditional and AI cheating methods. Cheating methods today are growingly stronger, making traditional anti-cheating methods less and less helpful. The CV AimBot, which does not include any illegal operation but only reading the game footage, has made all traditional anti-cheating methods de facto useless. We developed a CV AimBot from scratch and used it to generate the sample data. We gathered datasets from traditional AimBot and newly invented CV AimBot and run it through an RNN. Our RNN shows high accuracy against both traditional and CV AimBot. It also shows that introducing data from CV AimBot with significantly lower the accuracy of the result from RNN.