PhD Proposal: A Virtual Human Visual System for Visualization Design
Sungbok Shin
https://umd.zoom.us/my/sbshin90
Monday, August 28, 2023, 10:00 am-12:00 pm 
Abstract
A considerable volume of contemporary, creative, and complex visualizations are crafted by expert visualization designers. The primary objective of these designers is to effectively deliver their intended message to the target audience.
However, achieving the goal is not easy. The task often involves an iterative process of trial-and-error, which by its nature, is time-consuming. Access to a consistent and readily available source of feedback can help designers make decisions regarding their design choices. In practice, designers mainly rely on comments from their professional colleagues and peers, but this approach is not without drawbacks. First, they may not always provide objective feedback, because their comments and critique can influence interpersonal relationships with the designer. Second, these feedback sources are sometimes unavailable when required. Last but not least, alternative methods such as user studies, or crowdsourcing can be useful, but they come with additional burdens, such as increased costs and significant preparatory efforts.
I aim to address the drawbacks and instead provide objective, readily available, and pragmatic feedback for visualization designers by creating a virtual machine that delivers information about how the targeted audience would look at the visualization. To that end, I design a virtual human visual system called Perceptual Pat. Perceptual Pat is a system that consists of a suite of perceptual filters, where each filter is an imitation of how humans look at the visualizations (for example, eye movement, color vision deficiencies, etc.). My focus on a virtual human visual system further spans into two directions: (1) finding an optimal method to instill human perception into machines, and (2) exploring methods to effectively deliver perceptual feedback to visualization designers. To find how to optimally install human perception into machines, I created a Scanner Deeply, a framework to create the most optimal gaze prediction model for visualizations. I conducted a user study with visualization designers to understand how designers use the virtual human visual system. I also plan to conduct studies to explore effective methods to deliver perceptual feedback.
Examining Committee
Bio
Sungbok Shin is a 5th year Ph.D. Student in the dept. of Computer Science at the University of Maryland, College Park. He is a member of the VisTeam in Human-Computer Interaction Laboratory (HCIL), working with Dr. Niklas Elmqvist. He is interested in creating tools that imitate human perception to help visualization design. He received his Master's and Bachelor's degree in Computer Science and Engineering from Korea University, South Korea.
This talk is organized by Tom Hurst