Title
Human-Centered Dimensionality Reduction
Speaker
Hyeon Jeon, Ph.D. Student, Human-Computer Interaction Lab.,
Dept. of Computer Science and Engineering, Seoul National University
Abstract
Dimensionality reduction (DR) is one of the most widely used techniques to visualize and analyze high-dimensional data. However, the way we use DR in practice is not human-centered, i.e., hardly
reflects and supports users' needs. In this talk, I discuss recent efforts to improve the way we leverage DR for visualizations to align with diverse user needs. This is done by first redesigning how we evaluate and interact with DR projections to better meet
the needs of communicative visualization designers and data analysts. We also reflect on our current practice of using DR within visual analytics to inform the design of systems that effectively address target users’ needs. I conclude the talk by discussing
the importance of democratizing human-centered approaches to a broader range of machine learning models that affect visual analytics.
Bio
Hyeon Jeon (hyeonjeon.com) is a final-year Ph.D. Student at the Department of Computer Science and Engineering,
Seoul National University, Seoul, Korea. He aims to make machine learning more human-centered to enable reliable and efficient visual analytics. To this end, he develops machine learning algorithms, quality metrics, and interaction techniques for visual analytics.
His research was published in prestigious visualization, machine learning, and HCI venues, such as TVCG, VIS, PacificVIS, TPAMI, and CHI. He received a B.S. in Computer Science and Engineering from POSTECH, Pohang, Korea. His Ph.D. study is in part supported
by Google Ph.D. fellowship in Human-Computer Interaction and Visualization.