Dec 05, 2020 09:00:00

An AI that distinguishes cheating of online games from 'data traffic' is born

Artificial intelligence (AI) has been developed to distinguish '

cheat ', which is a fraudulent act that favors competitive games with an external program, from 'data traffic'.

GCI: A GPU Based Transfer Learning Approach for Detecting Cheats of Computer Game --IEEE Journals & Magazine
https://ieeexplore.ieee.org/document/9154512

Computer Scientists Create Ultimate Cheat-Detection System That Can Be Used in Any MMO Game | Tech Times
https://www.techtimes.com/articles/254269/20201118/computer-scientists-create-ultimate-cheat-detection-system-used-mmo-game.htm

Computer Scientists Launch Counteroffensive Against Video Game Cheaters --News Center --The University of Texas at Dallas
https://www.utdallas.edu/news/science-technology/stopping-video-game-cheaters-2020/

At the same time that competitive games such as e-sports have achieved remarkable growth, the cheat tools used in competitive games have also been developing, and the development of a cheat detection system as a countermeasure has not caught up. Under these circumstances, the computer scientist team at the University of Texas at Dallas focused on the 'data traffic' when the cheat tool operates, and developed an AI that 'identifies the cheat tool user from the data traffic'. did.

The research team conducted an experiment in which 20 subjects actually used the cheat tool on a dedicated server in the competitive FPS game 'Counter-Strike: Global Offensive ', which is also popular as an e-sports. 'Aimbot' that automatically aims at the enemy, 'Wallhack' that can make the wall transparent and know the position of the enemy, 'Speedhack' that sets the movement speed of your character to an unusual value 3 Track traffic between a dedicated server and a player in real time, with subjects using different types of cheat tools. In addition to the amount of incoming and outgoing packets, we measured packet size, transmission timing, packet continuity, and so on.

As a result of the analysis, when these cheat tools operate, a pattern peculiar to changes in data traffic was confirmed. By using this confirmed pattern for machine learning, the research team has developed 'AI that can predict whether cheat tools are used from the amount of data traffic'. The developed AI can determine the usage status of the cheat tool even when a large number of players participate.

Since the developed AI operates only on the original server operated by the research team, the research team plans to devise an approach to make AI correspond to the official server and secure the detection mechanism in the future. Muhammad Shihaburu Islam, the lead author of the treatise, said, 'It is also possible to learn the AI developed this time using our own data,' and told the game development company the AI that he developed. I called on you to use it.

'Our goal is to ensure that games like Counter-Strike are fun and fair to all players,' Islam said.