Research on Automatic Balance Patch Technology Through AI Analysis

Recently, the scale of the video game industry has become large, and several developers are rapidly producing and releasing games in accordance with the changes in the market. However, due to the lack of manpower, time, and budget of small developers, there is a problem that the game testing can not be done in depth. To resolve this problem, in this paper, we propose a model that balances the game using ML-Agent, which is a unity machine learning asset. We created AI models that took a lot of trial and error for accurate testing, and in addition, we improved the accuracy of the balance by reliably recording the win rate according to the number of characters.