100 likes | 248 Vues
This outline presents various sample projects from the Computational Intelligence course at Washburn University, focusing on AI applications in gaming. Students tackle challenges like informed search, optimization algorithms, adversarial search, and machine learning through engaging projects such as creating a predator-prey game, navigating corn mazes, solving Rubik's Cubes, and playing Tic-Tac-Toe. With practical exercises, students learn to implement algorithms like A*, LRTA*, MiniMax, and Genetic Algorithms, while also considering broader pedagogical approaches to foster understanding in diverse learning environments.
E N D
Applications for Gaming in AI Sample Projects from Computational Intelligence Course at Washburn University
Outline • Sample projects from this course • Challenges
Applications of Informed Search • Build Game Board where Predator is Searching a matrix looking for least cost path to Prey • Task Environment is fully observable • Both Single and Multi-Agent Implementations [i.e. both predator and prey are moving] • A* • Idea: avoid expanding paths that are already expensive • Evaluation function f(n) = g(n) + h(n)
Applications of Informed Search • Build a Corn Maze where agent finds its way through the maze • LRTA* • Used to solve problems where planning and action are interleaved and environment is safely-explorable • Search to find the optimal solution to a randomly selected scrambled Rubik's Cube • Iterative Deepening A* (IDA*)
Applications of Optimization Algorithms • N-Queens • Place n=8 queens on board with no attacking queens • Hill Climbing • Successor function generates 64 new boards • Pick the best new board • Beam Search - Pick best k moves • Genetic Algorithms • Successor function applies Fitness Function, Cross-Over, and Mutation to generate new population of moves
Applications of Adversarial Search • Tic-Tac-Toe • MiniMax [with Alpha-Beta Pruning] • Setting a cutoff where levels can be novice through Master Level • Mastermind • Please don’t ask me questions about this game… student is currently researching
Applications of Machine Learning • You enter how you would vote on a set of legislative bills and I [the computer] will predict your political party • Naïve Bayes • Guess your Cartoon Character based on the answer to twenty questions • Nearest Neighbor
Challenges • Understanding is not necessarily trivial • Significant career opportunities in emerging fields that are not just related to gaming • [e.g. Learning Science and Web Science]. • The challenge • Develop the proper pedagogy and scaffolding that will support student learning of these concepts. • Course needs to be adaptable to meet the needs of many types of students
References • [1] American Association for Artificial Intelligence, 2006, Games and Puzzles, http://www.aaai.org/AITopics/html/games.html, retrieved December 6, 2006 • [2] Russell S. and Norvig R., Artificial Intelligence a Modern Approach, 2ed., 2003, Pearson Education, Inc. • [3] Bourg D. M. and Seemann G., AI For Game Developers, 2004, O’Reilly Media, Inc