He gives a useful explanation of the value of Deep Learning in working on a problem like cheating in an online game with millions of monthly players. Deep Learning can build a function that maps inputs to outputs. Consider y=(f)x. With Deep Learning, f can be applied to never-before-seen x and get sensible y, which is pretty powerful, especially given that without Deep Learning, the job of solving f falls on engineers, who are always difficult to find, and probably prefer to avoid the “treadmill of work.”
We’re pretty familiar with the “treadmill of work,” as we’ve been in the online fraud space for several years. There are a number of similarities between the problem of cheating in gaming and perpetrating online fraud against eCommerce and financial institutions. That’s why we’re in the process of creating our own anti-cheating solutions for the mobile gaming and esports space. We call it Elympics. We recently finished the multiplatform SDK (software development kit) portion of the project. With the SDK, we can look at the attributes of a device as well as the history of a player’s activity in a given game, even their learning curve. The idea is to give game developers easy-to-implement libraries that will provide advanced matchmaking tools based on Machine Learning algorithms and an anti-doping mechanism. It’s in our production servers and we’re using it to analyze cheating scenarios from an RPG game that was created especially for the project by our project partner Daftmobile.
The SDK’s structure will consist of the following elements:
- Library for collecting and processing the data
- Matchmaking service (the process of connecting players together for online play sessions)
- Anti-doping service (fraudulent behavior of players)
- Web panel that will show relevant matchmaking and doping statistics (based on behavioural analysis, the player skills and game history)
The detection will be based on data coming from many sources, such as software, hardware, network and behaviour of players. Our intent is to create an advanced matchmaking mechanism that will be applicable for all kinds of mobile eSports games. Elympics will perform matchmaking and detect doping in the following three scenarios:
- real-time game player vs player
- real-time game in a multiplayer model
- turn-based game
Detecting “doping” is crucial to the success of matchmaking, as getting paired with cheaters is a really fast way to lose legitimate players in a game.
We hope that developers will use elympics to manage cheats in their games in a convenient, cost-effective manner since they will be built as components that can be attached to the most popular game development engines. Stay tuned for more updates.
If you wish to detect and protect your business against mobile game hacks and cheaters, we're here to help you with the perfect fraud prevention solution...