Overview:
- Implement a bench of algorithm to compare the performance on playing Super Mario.
- Implement and optimize Depth-first and Breadth-first search to find a perfect path.
- Implement Best-first search with a heuristic that gives an admissible estimate of the distance to goal state.
- Implement the evolutionary algorithm which include Evolution Strategy and Genetic Algorithm.
- Implement supervised learning algorithm(perceptron,decision tree).
- Implement Q-learning algorithm and come up with a way to simplify the implementation and make it can work for the path finding for game.
How to get it to run:
- Download the competition mario code from http://julian.togelius.com/mariocompetition2009/gettingstarted.php.
- Put my Mario AI into the src/competition/cig/yuxiao/ folder
- Launch \src\ch\idsia\scenarios\Play.java, change different agent by changing program argument to compare the performance.