Can we use a neural network that is trained using Reinforcement Learning for dynamic game level difficulty designing in realtime?

Question

I am a newbie to Machine Learning and AI. As per my understanding, with the use of reinforcement learning (reward/punishment environment), we can train a neural network to play a game. I would like to know, whether it possible to use this trained model for deciding the difficulty of the next game level dynamically in realtime according to a player's skill level? As an example, please consider a neural network is trained using Reinforcement Learning for playing a mobile game (chess/puzzle, etc.). The game is not consists of a previously designed static set of game levels. After the training, can this model use to detect a particular player's playing style(score, elapsed time) to dynamically decide the difficulty of the next game level and provide customized game levels for each player in realtime?

Thank you very much and any help will be greatly appreciated.

Answer 1

There have been many researches in dynamic difficulty adjustment (DDA). I see this one is quite explaining: AI for Dynamic Difficulty Adjustment in Games. However, there are many factors when we are trying to do dynamic difficulty adjustment. As explained in paper above, one major problem is it is sometimes hard to make sure the created model will still deliver the intended experience.

You can also read this paper about game design and DDA on ace05The Case for Dynamic Difficulty Adjustment in Games .

So for short, you can use neural network -- or other learning methods -- to do dynamic difficulty adjustment. But it's more about game design and experience impact of your DDA.