More about HKUST
Generative AI for Data Visualization and Analysis
Speaker: Dr. Aoyu WU PostDoc Fellow, Harvard University and CSE Best PhD Dissertation Award (2021-22) Recipient Hong Kong University of Science and Technology Title: "Generative AI for Data Visualization and Analysis" Host: Prof. Kai Chen Date: Monday, 8 May 2023 Time: 4:00pm - 5:00pm Venue: Lecture Theater F (Leung Yat Sing Lecture Theater) near lift 25/26, HKUST Abstract: In today's data-driven world, visualization plays a crucial role in analyzing and communicating information. Professionals ranging from business analysts to academic researchers need to explore and narrate data stories, but many encounter obstacles when creating custom, expressive visualizations. In this presentation, I will discuss a research initiative that utilizes AI techniques to efficiently generate and analyze data visualizations. Building upon concurrent research, I will delve into an investigation of large language models' (various versions of GPT) capabilities in generating data visualizations. Throughout the discussion, I will highlight novel and intriguing phenomena, as well as potential modes of human-LLM interaction observed during the experiment. To conclude, I will propose several research directions that aim to further democratize data visualization and analysis and discuss the implications of human-LLM interaction. ******************* Biography: Aoyu WU is a PostDoc Fellow at Harvard, advised by Prof. Fernanda Vias and Prof. Martin Wattenberg in the Insight and Interaction Lab. In addition, he is a teaching assistant at Harvard Business School. Aoyu's research is centered on human-computer interaction and data visualization. He is particularly invested in designing and evaluating intelligent, interactive systems that facilitate the creation of data visualizations for communication, data analysis, and data-driven decision-making. His work has garnered numerous industrial and academic accolades, including the Microsoft Research PhD Fellowship, Hong Kong ICT Award, Asia- Pacific ICT Award, and ACM CHI 2022 Honorable Mention, and has been featured in media outlets such as SCMP and Tai Kung Pao. His Ph.D. thesis addresses the growing public demand for access to and analysis of data. As data visualizations have become a primary medium for public data communication, the rapid increase in their use has led to challenges in readability and quality. Aoyu's thesis explores these issues through a comprehensive blend of research methods, including literature reviews, empirical studies, and machine learning techniques. He specifically concentrates on developing innovative recommendation systems for creating high-quality visualizations and formalizing visualizations as a new first-class object for efficient analysis. By combining these two approaches, his dissertation contributes to a novel online knowledge ecosystem that both extracts knowledge from web visualizations and aids the public in generating new visualizations to convey data.