Dimensionality reduction and visualization of geoscientific images via locally linear embedding

Authors:
Fabio Boschetti
Affiliations:
CSIRO Exploration & Mining, PO Box 1130, Bentley WA 6102, Australia
Venue:
Computers & Geosciences
Year:
2005

Citing 4
Cited 1

Laplacian Eigenmaps for dimensionality reduction and data representation

Neural Computation
Controlling and investigating cellular automaton behavior via interactive inversion and visualization of the search space

New Generation Computing
A Nonlinear Mapping for Data Structure Analysis

IEEE Transactions on Computers
Measure of similarity between geological sections accounting for subjective criteria

Computers & Geosciences

Charting of geoelectric potential signal dynamics via geometrical techniques and its possible relation to significant earthquakes in Western Greece

Computers & Geosciences

Quantified Score

Hi-index	0.00

Visualization

Abstract

The locally linear embedding (LLE) algorithm is useful for analyzing sets of very different geoscientific images, ranging from smooth potential field images, to sharp outputs from modeling fracturing and fluid flows via cellular automata, to hand sketches of geological sections. LLE maps the very high-dimensional space embedding the images into 2-D, arranging the images on a plane. This arrangement highlights basic relationships between the features contained in the images, thereby greatly simplifying the visual inspection of the entire data set. Other applications include image classification, and visualization of the results of inverse modeling of geological problems in order to characterize domains of different mechanical behavior.