A Survey on Data-Centric AI: Tabular Learning from Reinforcement Learning and Generative AI Perspective

• DOI: 10.48550/arxiv.2502.08828
• Publication Date: 2025-02-12
• Authors: Wangyang Ying; Cong Wei; Nanxu Gong; Xinyuan Wang; Haoyue Bai; Arun Vignesh Mala...; Sixun Dong; Dongjie Wang; Denghui Zhang; Yanjie Fu
• Keywords: Generative model

Tabular data is one of the most widely used formats across various domains such as bioinformatics, healthcare, and marketing. As artificial intelligence moves towards a data-centric perspective, improving quality essential for enhancing model performance in tabular data-driven applications. This survey focuses on optimization, specifically exploring reinforcement learning (RL) generative approaches feature selection generation fundamental techniques refining spaces. Feature aims to identify retain informative attributes, while constructs new features better capture complex patterns. We systematically review existing methods engineering, analyzing their latest advancements, real-world applications, respective strengths limitations. emphasizes how RL-based contribute automation engineering. Finally, we summarize challenges discuss future research directions, aiming provide insights that drive continued innovation this field.