The AI agent has two main methods of coming up with a good move. When processing time and RAM space is sufficient, it utilizes a Alpha Beta Pruning Search Tree algorithm to search the seed amount that it will have after few rounds of moves for all current legal moves; while running time and RAM space is limited or board states are numerous to be processed, it will filter and compare the pre-defined heuristics, evaluate the board state at a level and sum up all heuristic factors to find the best move. A search tree algorithm is more accurate than Heuristic method because it inspects all cases in the following “n” rounds instead of narrowing itself down to one certain state. Since the given constrains give enough space and running time for the AI agent to go for at least a depth of 6 rounds for alpha pruning search, and this Has game has a maximum of 32 pits for one side to be counted for each turn, it will use Alpha Beta Pruning mainly.
Equations: EV(expected value)=P(outcome 1)+(1-P)(outcome 2) EU(expected utility)=P(√(outcome 1))+(1-P)(√(outcome 2))
Metanautix now has Personal Quest online where individual users can download and do analytics on desktop. I have been working on a new systematic way to learn music theory and do musical analysis with mathematical matrix and vectors, and Quest is a critical tool in query on large dataset such as a pool of thousands of song scores. In this article, I will talk about my methodology in detail.
An overview for my system will be: