Conclusions and Reflections - leungjunbob/AIP-project GitHub Wiki
In this project, our team conducted a comprehensive and in-depth analysis and implementation of AI strategies for the Splendor game. We meticulously analyzed the problem and selected appropriate techniques for implementation. Specifically, we implemented three technologies: Minimax, SARSA, and MCTS. All three technologies achieved commendable results. Among them, Minimax demonstrated the best performance, achieving a 72.5% winning rate (out of 40 games) on the server. In local comparisons, Minimax secured a 90% winning rate (out of 20 games) against SARSA and a 95% winning rate (out of 20 games) against MCTS.
Our reports provide comprehensive coverage of all critical aspects of the project. From systematically analysing the problem to providing solid reasons for technology choices, every step is rigorously built on logic and data analysis. During the implementation process, we not only elaborated on the specific reasons for adopting each technology, but also successfully demonstrated the practical effects and potential advantages and disadvantages of these technologies. Especially in terms of problem models and implementation, we demonstrated the complexity and sophistication of the AI decision-making process through clear logic and detailed technical details. At the same time, we solved the challenges faced in the implementation process of multiple technologies. The solution of these challenges not only enhanced the functions of our system, but also provided valuable experience for future optimisation and improvement.
In addition, through experimental design and result presentation, we ensured the reliability of the research results. At the same time, it also shows the comparison of the process from the BFS algorithm initially completed in the personal part to the final successful implementation of the three technologies. In summary, this project demonstrates our team’s capabilities in technical implementation and problem solving, as well as our thoughtful consideration of future improvements. Our work is a useful addition to the field, and going forward, we look forward to further deepening and extending our research on this basis.