Neural networks can accurately identify cheaters

American students at the University of New Mexico have been able to prove that neural networks are able to detect additional software for shooting CS: GO with an accuracy of 98%. Throughout the scientific work, a lot of tests were done using the Multilayer perceptron and Learning vector quantization neural networks. They worked together with demos of regular players and cheaters. A significant effect was shown by the mathematical model MLP, the error of which was only 2%.

At the well-known forum Reddit, in which this scientific work was published, a Valve employee was noted under the nickname vMcJohn. He noted that the developed anti-cheat VAC with artificial intelligence will work in the same way, that is, conduct an analysis of the player's behavior during the game. To release such an anti-cheat, you need an accuracy of 100%.