Detecting Cheating in Computer Games using Data Mining Methods

Abstract

Cheating is prevalent in the gaming industry and is causing several types of uproars. Counter-Strike is a popular franchise which has had up to peaks of 850 thousand players at once playing. Professional events such as ESL One Cologne 2015 had 1.3 million concurrent viewers. Video games are a huge industry, sadly, integrity of the players is lacking. Professional players have been found to be using cheats at events, and cheaters in casual matches are expected to be found regularly. Current anti-cheat systems may use signature-based approaches and even heuristics, but none of them have explicitly stated to be using data mining techniques. Although signature methods may catch a lot of not technically adept cheaters, the ones we need worry about are the ones using metamorphic or polymorphic cheats that stay undetectable to these systems. To detect these kinds of cheats, another method needs to be put in place. Data mining techniques used for detecting zero-day malware as well as player behaviorbased techniques for detecting cheaters are discussed in this proposal. The results of these methods are promising and will hopefully rid the gaming industry of ill-minded players.