Data-Driven Computer Go Based on Hadoop
Data-Driven Computer Go Based on Hadoop.Although the research for computer-based Go AI engines started at about 1980’s, Go is currently considered as the last board game which need to be implemented using computer systems. Most current Go implementations are using tree-search technique to find the best moves, however it is nearly impossible to review all possible candidate moves in Go games.
In this paper, we propose a totally different approach: a data-driven Go engine. Hugh amount of game log database is stored in Hadoop servers which support parallel pattern retrieval. In each game status, Go engine transfers the board configuration to Hadoop server to find the best move from database. We anticipate that this simple data-driven approach could be expandable to the one having learning capability and could be easily merged with other traditional approaches for better performance.
Similar IEEE Project Titles
- Translation Memory for a Machine Translation System Using the Hadoop Framework
- Signature based malware detection for unstructured data in Hadoop
- Adaptive Task Scheduling Strategy Based on Dynamic Workload Adjustment for Heterogeneous Hadoop Clusters
- Performance Characterization of Hadoop and Data MPI Based on Amdahl’s Second Law
- PigOut: Making multiple Hadoop clusters work together