Dimensionality reduction of multidimensional temporal data through regression

Authors:
Lalitha Rangarajan;P. Nagabhushan
Affiliations:
Department of Studies in Computer Science, University of Mysore, Manasagangothri, Mysore 570 006, Karnataka, India;Department of Studies in Computer Science, University of Mysore, Manasagangothri, Mysore 570 006, Karnataka, India
Venue:
Pattern Recognition Letters
Year:
2004

Citing 3
Cited 1

Optimum Uniform Piecewise Linear Approximation of Planar Curves

IEEE Transactions on Pattern Analysis and Machine Intelligence
Algorithms for clustering data

Algorithms for clustering data
Dynamic strip algorithm in curve fitting

Computer Vision, Graphics, and Image Processing

Linear regression for dimensionality reduction and classification of multi dimensional data

PReMI'05 Proceedings of the First international conference on Pattern Recognition and Machine Intelligence

Quantified Score

Hi-index	0.10

Visualization

Abstract

A new method for pattern classification of multidimensional temporal data/images is proposed. In temporal data/ images, each feature of a sample/pixel is not just a single numerical value, but a set (vector) of real values. The method proposed transforms the pattern of change in the feature values, over time, into representative patterns, termed as symbolic objects [Bock, Diday (Eds.), Analysis of Symbolic Data, Springer Verlag, 2000], which are obtained through regression lines. Since a regression line symbolizes a sequence of numerical values of a feature vector, the so defined symbolic object accomplishes dimensionality reduction of the temporal data. A new distance measure is devised to measure the distances between the symbolic objects (fitted regression lines) and clustering is performed. The method is very versatile and is readily applicable to any multidimensional temporal image. The algorithm is tested on two different data sets.