A Survey on Deep Reinforcement Learning for Data Processing and Analytics

Data processing and analytics are fundamental and pervasive. Algorithms play\na vital role in data processing and analytics where many algorithm designs have\nincorporated heuristics and general rules from human knowledge and experience\nto improve their effectiveness. Recently, reinforcement learning, deep\nreinforcement learning (DRL) in particular, is increasingly explored and\nexploited in many areas because it can learn better strategies in complicated\nenvironments it is interacting with than statically designed algorithms.\nMotivated by this trend, we provide a comprehensive review of recent works\nfocusing on utilizing DRL to improve data processing and analytics. First, we\npresent an introduction to key concepts, theories, and methods in DRL. Next, we\ndiscuss DRL deployment on database systems, facilitating data processing and\nanalytics in various aspects, including data organization, scheduling, tuning,\nand indexing. Then, we survey the application of DRL in data processing and\nanalytics, ranging from data preparation, natural language processing to\nhealthcare, fintech, etc. Finally, we discuss important open challenges and\nfuture research directions of using DRL in data processing and analytics.\n