Discriminatory Data Mapping by Matrix-Based Supervised Learning Metrics

Authors:
M. Strickert;P. Schneider;J. Keilwagen;T. Villmann;M. Biehl;B. Hammer
Affiliations:
Leibniz Institute of Plant Genetics and Crop Plant Research Gatersleben,;Institute for Mathematics and Computing Science, University of Groningen,;Leibniz Institute of Plant Genetics and Crop Plant Research Gatersleben,;Research group Computational Intelligence, University of Leipzig,;Institute for Mathematics and Computing Science, University of Groningen,;Institute of Computer Science, Technical University of Clausthal,
Venue:
ANNPR '08 Proceedings of the 3rd IAPR workshop on Artificial Neural Networks in Pattern Recognition
Year:
2008

Citing 8
Cited 1

Independent component analysis: algorithms and applications

Neural Networks
Generalized relevance learning vector quantization

Neural Networks - New developments in self-organizing maps
Supervised Neural Gas with General Similarity Measure

Neural Processing Letters
Targeted projection pursuit for visualizing gene expression data classifications

Bioinformatics
Feature Extraction: Foundations and Applications (Studies in Fuzziness and Soft Computing)

Feature Extraction: Foundations and Applications (Studies in Fuzziness and Soft Computing)
Iterative RELIEF for Feature Weighting: Algorithms, Theories, and Applications

IEEE Transactions on Pattern Analysis and Machine Intelligence
Interactive and Dynamic Graphics for Data Analysis With R and GGobi

Interactive and Dynamic Graphics for Data Analysis With R and GGobi
Generalized relevance LVQ (GRLVQ) with correlation measures for gene expression analysis

Neurocomputing

Features and Metric from a Classifier Improve Visualizations with Dimension Reduction

ICANN '09 Proceedings of the 19th International Conference on Artificial Neural Networks: Part II

Quantified Score

Hi-index	0.00

Visualization

Abstract

Supervised attribute relevance detection using cross-comparisons (SARDUX), a recently proposed method for data-driven metric learning, is extended from dimension-weighted Minkowski distances to metrics induced by a data transformation matrix 茂戮驴for modeling mutual attribute dependence. Given class labels, parameters of 茂戮驴are adapted in such a manner that the inter-class distances are maximized, while the intra-class distances get minimized. This results in an approach similar to Fisher's linear discriminant analysis (LDA), however, the involved distance matrix gets optimized, and it can be finally utilized for generating discriminatory data mappings that outperform projection pursuit methods with LDA index. The power of matrix-based metric optimization is demonstrated for spectrum data and for cancer gene expression data.