Title:
Investigating the Applications of Machine Learning in the Context of a 
Turn-Based Strategy Game Through a Computer AI System.

Objective:

To evaluate the effectiveness of a learning AI system against a Human opponent 
in strategic thinking.
Note: There are other smaller research applications that can be applied in 
the same context, so while Sam and I will primarily be focusing on a machine 
AI we can also undertake smaller seperate research areas, such as 
machine-human compatability and interfacing.

Justification:

Strategy games are a popular source of entertainment and constantly strive for 
a more fun and competitive AI component. Frequently humans find a computer AI 
opponent to make for a boring match. As of yet, there has never been an AI 
strategy system comparable to human strategy, primarily because humans can 
understand the trends in the AI, but a human is much more difficult to see 
trends in. Therefore, a learning AI system would be a breakthrough in the 
modern gaming experience.

Description:

Using a self-created strategy game based off a popular strategy game, 
Civilization II, my partner and I will create and playtest an AI system. We 
will rate the AI system based on the AI's ability to formulate long-term 
strategies and effectively execute them to best target weaknesses in the human 
player. The final test will be play testing the game on a randomly selected 
audience and evaluating the AI's effectiveness as described above and by 
player comments.

Limitations:

The evaluation system is somewhat arbitary, largely based on opinions. The 
performace and speed of the game must remain within playable limits while 
running an advanced AI. The test for the AI may be very time consuming if the 
AI choose to play too defensively (there must be a balance between 
effectiveness and aggressiveness in the AI, defensive playing could make the 
game boring). The AI must be effective but also fun to play against.