Drexel University

Dealing with Adversarial Player Strategies in the Neural Network Game iNNk through Ensemble Learning

Neural network-based games frequently face stability challenges when players identify and exploit specific model weaknesses to disrupt intended gameplay mechanics.

• An ensemble learning framework combining a baseline neural network with specialized transfer-learning models effectively mitigates player exploitation of AI weaknesses in games like iNNk.
• The dual-layered architecture maintains high accuracy across diverse player inputs, such as distraction tactics and non-standard line work, without degrading performance in standard gameplay.
• Specialized components trained on targeted adversarial datasets prevent the model from overfitting on limited data while ensuring consistent AI behavior.

Player-Centered AI for Automatic Game Personalization: Open Problems

Game personalization research must transition from a system-centered focus on algorithmic complexity toward a player-centered framework that prioritizes individual cognitive processes and user needs.

Reflection in Game-Based Learning: A Survey of Programming Games

Educational programming games frequently prioritize evaluative accuracy over the cognitive processes required for deep learning.