An Empirical Study of Machine Learning Algorithms Applied to Modeling Player Behavior in a "First Person Shooter" Video GameUniversity of Wisconsin--Madison, 2002 - 46 pages |
Expressions et termes fréquents
Accelerate acceptable accuracy achieve lower actions Actual agent allow applied artificial neural network Backward bagging Baseline basic began behavior better boosting classifiers closest enemy ClosestGoal coded collected combat Computer considered created CurrentMove data set decision tree determined developers Distance domain dropped Enemies in Sector EnemySector ensemble methods expert player Face Direction feature set Figure function goal Health important increase independence input instances Jump known learner learning algorithms lower error rates machine learning algorithms Matrices means Move Direction move forward Naïve Bayes necessary needed negative neural network ANN node Number Enemies output pair performance person shooter possible Predicted produced recent recorded represent rule sample Set Error Rate shown shows situation sizes standard step Table task Test Set Error thesis training examples training set tune set values video game weights