Type
Text
Type
Dissertation
Advisor
Mueller, Klaus | Kaufman, Arie | Gu, Xianfeng | Grossberg, Michael.
Date
2016-12-01
Keywords
Computer science | depth cue, high dimensional data, information visualization, trackball, visual analytics
Department
Department of Computer Science
Language
en_US
Source
This work is sponsored by the Stony Brook University Graduate School in compliance with the requirements for completion of degree.
Identifier
http://hdl.handle.net/11401/77256
Publisher
The Graduate School, Stony Brook University: Stony Brook, NY.
Format
application/pdf
Abstract
Data with many attributes have become commonplace in a wide range of domains. In these data, the most interesting relations are often multivariate and are generally confusing to most people. Efforts have been made to design proper tools to recognize those high dimensional relationships reliably but those tools are often far off from making use of the innate 3D scene understanding capabilities of the human visual system. We present a framework that eases this barrier by design, called the Subspace Voyager. It decomposes a high-dimensional data space into a continuum of generalized 3D subspaces. Analysts can then explore these 3D subspaces individually via the familiar trackball interface and use additional facilities to smoothly transition to adjacent subspaces for expanded space comprehension. On top of the Subspace Voyager, we propose a novel 3D shaded shape representation for non-spatial data. This representation visualizes data matrices in the most natural 3D forms that include depth cues, such as occlusion, shading, perspective distortion, shadows, and so on. Our user study suggests that mainstream users prefer shaded displays over scatterplots for visual cluster analysis tasks after receiving training for both. And further, our experiments also provide evidence that 3D displays can better communicate spatial relationships, size, and shape of clusters. When designing those tools, we often had difficulties acquiring proper testing data. We therefore propose an interactive data generation tool – SketchPadN-D. The core concept in our SketchPadN-D is WYSIWYG (What You See Is What You Get) because it allows users to generate dataset in the same interface they use to visualize it such that they do not need to switch back and forth between data manipulation and visualization tools. | Data with many attributes have become commonplace in a wide range of domains. In these data, the most interesting relations are often multivariate and are generally confusing to most people. Efforts have been made to design proper tools to recognize those high dimensional relationships reliably but those tools are often far off from making use of the innate 3D scene understanding capabilities of the human visual system. We present a framework that eases this barrier by design, called the Subspace Voyager. It decomposes a high-dimensional data space into a continuum of generalized 3D subspaces. Analysts can then explore these 3D subspaces individually via the familiar trackball interface and use additional facilities to smoothly transition to adjacent subspaces for expanded space comprehension. On top of the Subspace Voyager, we propose a novel 3D shaded shape representation for non-spatial data. This representation visualizes data matrices in the most natural 3D forms that include depth cues, such as occlusion, shading, perspective distortion, shadows, and so on. Our user study suggests that mainstream users prefer shaded displays over scatterplots for visual cluster analysis tasks after receiving training for both. And further, our experiments also provide evidence that 3D displays can better communicate spatial relationships, size, and shape of clusters. When designing those tools, we often had difficulties acquiring proper testing data. We therefore propose an interactive data generation tool – SketchPadN-D. The core concept in our SketchPadN-D is WYSIWYG (What You See Is What You Get) because it allows users to generate dataset in the same interface they use to visualize it such that they do not need to switch back and forth between data manipulation and visualization tools. | 105 pages
Recommended Citation
Wang, Bing, "Innovations in High Dimensional Data Exploration, Representation and Generation" (2016). Stony Brook Theses and Dissertations Collection, 2006-2020 (closed to submissions). 3079.
https://commons.library.stonybrook.edu/stony-brook-theses-and-dissertations-collection/3079