9:00-10:40 Visualization Meets AI: Session 1
Opening and Invited Talk by Hanqi Guo on “Intelligent Visualization for Science”
The task of data visualization generally involves a design step, which requires the knowledge of the data domain and visualization methods to do well. Because of the immense design space for optimization, it can take both novices and experts substantial effort to derive desired visualization results from data for exploration or communication. Following the resurgence of artificial intelligence (AI) technology in recent years, in the field of visualization, there is a growing interest and opportunity in applying AI to perform data transformation and to assist the generation of visualization, aiming to strike a balance between cost and quality. The use of visualization to enhance AI is the other active line of research. This workshop, held in conjunction with IEEE PacificVis 2022, aims at exploring this emerging area of research and practice by fostering communication between visualization researchers and practitioners. Attendees will be introduced to the latest and greatest research innovations in AI-enhanced visualization (AI4VIS) as well as visualization-enhanced AI (VIS4AI), and also learn about further research opportunities. The workshop will be composed of paper presentations and invited talks.
Session 1: AI for VIS Representation and Prediction (Chair: Junpeng Wang)
9:00-10:00: Opening and Invited Talk: Hanqi Guo
- Intelligent Visualization for Science
- Today’s and future supercomputers enable scientists to accomplish large-scale simulations for earth system science, cosmology, and fluid dynamics and produce data at an unimaginable scale than previously possible. However, scientific data’s ever-increasing volume and complexity pose grand challenges to visualization and data understanding. At the same time, the rapid development of artificial intelligence (AI) necessitates new paradigms to incorporate AI in visualization workflows to accelerate scientific discoveries.
This talk will narrate three directions of using AI—intelligent infrastructure, filters, and surrogates—in scientific visualization workflows with our recent research. First, we build intelligent infrastructures to accelerate visualization algorithms. I’ll demonstrate how online-learning models could improve I/O performance and scalability of particle tracing for visualizing large-scale fluid flow data. Second, we innovate intelligent filters to substitute visualization algorithms. I’ll exemplify how deep neural networks excel in feature extraction and tracking in applications such as fusion plasma simulations. Third, we create intelligent surrogates to replace expensive visualization pipelines. I’ll present how surrogate models can synthesize and represent data and visualization results. I will summarize our multidisciplinary experiences researching intelligent visualization for sciences at the end of this talk.
- Dr. Hanqi Guo is a Computer Scientist in the Mathematics and Computer Science (MCS) Division of Argonne National Laboratory. He combines a unique background in visualization, high-performance computing, and artificial intelligence to lead several multidisciplinary DOE- and NSF-sponsored projects involving physicists, computer scientists, and mathematicians to address grand data challenges in computer science and domain sciences. Dr. Guo has published more than 50 papers and received the Best Paper Award in IEEE VIS 2019, Best Paper Award in IEEE PacificVis 2021, and several best paper honorable mentions in premier visualization conferences. Before starting his staff position in 2017, he received the Postdoctoral Performance Award in Basic Research in Argonne National Laboratory.
10:00-10:40: Paper Presentations
Session 2: Design and Evaluation of VIS-Assisted AI (Chair: Takanori Fujiwara)
11:10-12:10: Invited Talk: Yong Wang
- Visualization Meets AI: Automated Visualization Design and Evaluation
- Data visualization has been applied to facilitate data exploration in various applications. However, the design and evaluation of visualizations still require lots of manual effort. Existing visualization creation tools and packages (e.g., Tableau and D3) often require tedious manual specifications or programming, and the evaluation of visualizations often relies on subjective evaluations like user studies. It is still not an easy task to quickly create and evaluate visualizations, especially for non-expert users. The rapid development of artificial intelligence (AI) and the accumulation of visualization datasets have made it possible to address these issues by leveraging AI techniques. In this talk, I will briefly introduce our recent work on AI-powered data visualization:
a) AI for visualization design: how can AI techniques facilitate an end-to-end and explainable visualization generation/recommendation?
b) AI for visualization evaluation: how can AI techniques enable automated evaluation of visualization quality and similarity?
This talk ends with a discussion on future research opportunities.
- Dr. Yong Wang is a tenure-track assistant professor at the School of Computing and Information Systems, Singapore Management University (SMU). His research interests include information visualization, visual analytics and machine learning. His work has been published at premier venues in visualization and human-computer interaction, such as IEEE VIS, IEEE TVCG, and ACM SIGCHI. He received multiple paper awards, including the Best Paper Honorable Mention Award at IEEE VIS 2021, Best Poster Award at IEEE VIS 2019 and Best Paper Award at ACM IUI 2017. He also has served as a program committee for PacificVis, IEEE VIS, CIKM, and a session chair and program committee for IUI. Prior to joining SMU, he obtained his Ph.D degree from The Hong Kong University of Science and Technology. For more details, please refer to http://yong-wang.org/ .