Relationship Space Visualizing the Relationship of Multidimensional Data

  • Slides: 1
Download presentation
Relationship Space Visualizing the Relationship of Multidimensional Data in 2 D Relationship Space The

Relationship Space Visualizing the Relationship of Multidimensional Data in 2 D Relationship Space The properties of the relationship space depend on the different relations. Lei Wang and Arie Kaufman Computer Science Department, Stony Brook University Possible Applications: High Number of data Low v. Data Analysis: Visualize the relationship between data and features. v. Data Classification: Classify the data according to the relationship. v. Data Feature Detection: Extract features with special relation properties. v. Data Visualization: Visualize the data based on the relationship. v. Data Registration: Register the data with similar relationship in two data sets. High Strength of relation Low Volume Data Analysis Volume Data Visualization The relationship space is calculated from the features of the 3 D volume data, using feature distance as a relation. The relationship of the space is used for transfer function designation. The relationship space is invariant under noise, bias, and distortion. It shows the relationship of the data. The intensity shows the number of data under each relation strength. The spatial relationship of the grids encodes the relationship of data points Bonsai Volume Data Classification Mapping Between 2 D Data Space and 2 D Relationship Space The classification is obtained by classifying the grids in the relationship space. Volume Data Relationship Lobster Noise Bias Distortion Feature Relationship We use 2 D data for the convenience of display. The algorithm and properties are same in high-dimensional data space.