PROJECT TITLE :

Database Meets Artificial Intelligence: A Survey

ABSTRACT:

Both database management and artificial intelligence (AI) have something that can be learned from the other. On the one hand, AI has the potential to improve the intelligence of databases (AI4DB). For instance, traditional empirical database optimization techniques (such as cost estimation, join order selection, knob tuning, index and view selection) are not capable of meeting the high-performance requirements for large-scale database instances, various applications, and diversified users, especially when implemented on the cloud. The good news is that strategies based on learning can help alleviate this problem. On the other hand, database strategies have the potential to enhance AI models (DB4AI). For instance, artificial intelligence (AI) is difficult to implement in practical applications because it requires programmers to write complicated code and train complicated models. The complexity of using AI models can be simplified by employing database techniques, which can also be used to accelerate AI algorithm execution and provide AI capability within databases. Therefore, both DB4AI and AI4DB have been the subject of significant research as of late. In this article, we take a look at previous research pertaining to AI4DB and DB4AI. Techniques on learning-based configuration tuning, optimizer, index/view advisor, and security are discussed in this section pertaining to AI4DB. Reviewing AI-oriented declarative languages, AI-oriented data governance, training acceleration, and inference acceleration are some of the things we do for DB4AI. In conclusion, we discuss the challenges and opportunities for future research.