Dynamic difficulty adjustment for serious game using modified evolutionary algorithm

Dynamic Difficulty Adjustment (DDA) seeks to adapt the challenge a game poses to a human player. When the game is too easy the player can become bored, when it is too hard - frustrated. In the case of a serious game (educational game), additionally, without a balance between the player competence and the game challenge the game could repeatedly exploit the developed skills, or fail to achieve the pedagogical goals. In this paper evolutionary algorithm (EA) is used to find game settings suitable for the player of a serious math game. To reduce the number of training data and accelerate the search for the ‘right’ game difficulty level EA modifications are introduced. Various experiments are performed. The obtained results show that proposed methods can substantially decrease the time a human player has to wait for a suitable game level.