CIS6930-Project1-Tic-Tac-Toe

• This poject try to train AI with different algorithms to play Tic Tac Toe.

Availabe Algorithms

• Classification

1. K-nearest neighbors algorithm (KNN)
2. Support Vector Mahcine with Linear kernel (Linear-SVM)
3. Support Vector Mahcine with RBF kernel (RBF-SVM)
4. Multilayer perceptron (MLP)
5. Multilayer perceptron trained parameters by genetic algorithm (GA-MLP)

• Regression

1. KNN
2. Linear regression
3. MLP

Programs:

1. Classifiers.py: classifier program for all classification models
2. Regressors.py: regressor program for all regression models
3. PlayTicTacToe.py: play Tic Tac Toe with all trained models.
4. GA_ClfMLP_singleLabel.py: implement genetic algorithms to find best MLP parameters.

Folders:

1. TrainedCVresults: all trained CV records with different parameters.
2. TrainedParameters: all trained models' parameters.
3. RecordConfusionMatrix&Accuracy: all trained models' accuracy and confusion matrix.
4. GameParameters: trained models' with the best parameters for playing Tic Tac Toe.

Archives:

1. ClassifiersRes4AllModels.xslx: all trained models' training accuracy and testing accuracy.
2. RegressorsRes4AllModels.xslx: all trained models' training accuracy and testing accuracy.
3. Noise_ClassifiersRes4AllModels.xslx: all trained models' training accuracy and testing accuracy with adding noise.
4. Reduce9by10_ClassifiersRes4AllModels.xslx: all trained models' training accuracy and testing accuracy with reducing dataset.
5. GA_ClfMLP_singleLabelResults.xlsx: Results of GA-MLP classification model for single model.

Notes

• Please run the PlayTicTacToe.py program in 64 bits version of python. (Important.)
• Otherwise, the errors will be occured like (ValueError: Buffer dtype mismatch...)