Input Feature Extraction for Multilayered Perceptrons Using Supervised Principal Component Analysis

Authors:
Stavros J. Perantonis;Vassilis Virvilis
Affiliations:
Institute of Informatics and Telecommunications, National Center for Scientific Research ’’Demokritos‘‘, 153 10 Aghia Paraskevi, Athens, Greece, e-mail: sper@iit.demokrit ...;Institute of Informatics and Telecommunications, National Center for Scientific Research ’’Demokritos‘‘, 153 10 Aghia Paraskevi, Athens, Greece, e-mail: sper@iit.demokrit ...
Venue:
Neural Processing Letters
Year:
1999

Citing 4
Cited 2

Kendall's advanced theory of statistics

Kendall's advanced theory of statistics
Multi-layered feedforward neural networks for image segmentation

Multi-layered feedforward neural networks for image segmentation
An efficient constrained learning algorithm with momentum acceleration

Neural Networks
Characterization of the Sonar Signals Benchmark

Neural Processing Letters

The Minimum Number of Errors in the N-Parity and its Solution with an Incremental Neural Network

Neural Processing Letters
Feature selection via Boolean independent component analysis

Information Sciences: an International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

A method is proposed for constructing salient features from aset of features that are given as input to a feedforward neural networkused for supervised learning.Combinations of the original features are formed that maximize thesensitivity of the network‘s outputs with respect to variations ofits inputs. The method exhibits some similarity to Principal ComponentAnalysis, but also takes into account supervised character of the learning task. It is applied to classification problems leadingto improved generalization ability originating from the alleviationof the curse of dimensionality problem.