Visualizing High-Dimensional Graph Embeddings via Informed Multi-View Projections
Authors
Ya Ji (Northeastern University), Xuefeng Li (Northeastern University), Timo Brand (Technical University of Munich), Jacob Miller (Technical University Munich), Peng Zhang (Northeastern University), Stephen Kobourov (Technical University Munich), Yifan Hu (Northeastern University)
Presentation
- Session
- I feel tangled in a net
- Time
- Wednesday, Nov 11, 08:12 – 08:24 (US/Eastern) · session 08:00 – 09:30
- Location
- Hall America center
Links
Sign in to access the preprint PDF.
Sign in- Download Supplemental Material
Keywords
Graph visualization, dimensionality reduction.
Abstract
Graphs are commonly visualized in 2D, where humans readily interpret spatial relationships, yet such layouts often distort higher-dimensional structure. We propose to embed graphs in high-dimensional space and search for informative 2D viewpoints that optimize aesthetic and readability metrics (e.g., edge crossings and angular resolution), enabled by a novel differentiable surrogate for edge crossings. Numerical experiments show that these viewpoints consistently outperform standard 2D layouts, and can even surpass methods explicitly designed to optimize these metrics. We further introduce DataFly, an interactive system for exploring multiple candidate viewpoints through seamless navigation. A usability study demonstrates that our approach reveals structural patterns that remain hidden in conventional 2D visualizations.
For Practitioners
Data scientists, network analysts, and visualization practitioners working with graph-structured data would be interested in this paper. They could use our approach to explore multiple informative views of a graph, select projections that improve specific readability metrics, and reveal structural patterns that may be hidden in a single 2D layout.