题目：Data-Centric AI: Refining Representation Space from Decision-Making to Sequential-Generation Perspectives

主讲人：美国中佛罗里达大学 王东杰博士

主持人：统计学院 黄雁勇教授

时间：2023年8月1日（周二）上午10:00-11:00

地点：西南财经大学柳林校区诚正楼统计学院会议室

报告摘要：

In the realm of artificial intelligence, the quest for optimal representation space has been a critical pursuit, as it directly impacts the performance of various learning tasks. In this presentation, we delve into the concept of data-centric AI and explore two distinct approaches to refine the feature space: feature selection learning and feature generation learning. Approaching this study from two fundamental perspectives – the decision-making and sequential generation perspectives – we aim to shed light on the efficacy of each technique in enhancing the representation space. Through extensive evaluation of experimental results, it becomes evident that both feature selection learning and feature generation learning exhibit promising capabilities in shaping the representation space. This transformative capacity directly influences the performance of AI models, ultimately leading to superior results in various applications. By addressing the core tenets of data-centric AI, this research strives to highlight the crucial role played by feature selection and feature generation learning in refining the representation space. Armed with this knowledge, researchers and practitioners can make informed decisions when choosing the most suitable technique for specific AI tasks, ultimately advancing the field of artificial intelligence and its practical applications.

主讲人简介：

王东杰，美国中佛罗里达大学博士。他的主要研究方向是数据挖掘和机器学习，特别关注自动化数据科学系统在大规模数据问题上的应用，如智能城市、异常检测、根本原因分析、自动城市规划和用户行为分析等。他在顶级数据挖掘和人工智能会议和期刊上发表了25篇以上的论文，如TKDE、SIGKDD、AAAI和KAIS。他的研究成果分别取得了SIGSPATIAL 2020和ICDM 2021的最佳论文亚军奖。