Match-3 Level Similarity Analysis

A Python3 project for generating a similarity score for any 2 levels of a Match-3 game. This project includes 2 approaches as possible solutions: one method uses unsupervised learning with the help of a Simple AutoEncoder which was developed using PyTorch, and the other method is a rule-based algorithm based on the aggregation of level data for vectorization.

Dependencies

The dependencies of the project are available in requirements.txt in the root of the project. They can be installed with the following command:

pip install -r requirements.txt

Running the project

Run the project using the following command:

python main.py [algorithm]

where algorithm can take up the following values:

1. Train the model
2. Generate dense representation and level data
3. Generate scores for all levels
4. Optimize the weights for the vectors
5. Generate the plots
6. Rule-based algorithm score generation

Options 1-5 are for the AutoEncoder and option 6 is for the Rule-based algorithm

Note:

The pre-trained model is not available in this repository, but the model can easily trained and saved. Just execute python main.py 1 after installing the dependencies and follow the instructions to save the model. The model is saved and accessed by default at the path ./output/autoencoder/model.pth

Report

• The report for this project can be accessed here

Built With

• PyTorch - The ML framework used

Author

• Salil Deshpande - Master of Information Systems Management Student at Carnegie Mellon University, USA - Salild1011

License

This project is licensed under the Apache 2.0 License - see the LICENSE.md file for details

Acknowledgments

This project was given the right direction and was developed under the guidance of Mr. Simon Cheng Liu and Dr. Guo Xianghao of Levelup AI, Beijing.