Abstract
The projection maps and derived classification accuracies of a neural network (NN) implementation of Sammon's mapping, an auto-associative NN (AANN) and a multilayer perceptron (MLP) feature extractor are compared with those of the conventional principal component analysis (PCA). Tested on five real-world database, the MLP provides the highest classification accuracy at the cost of deforming the data structure, whereas the linear models preserve the structure but usually with inferior accuracy.
Original language | English |
---|---|
Pages (from-to) | 7-14 |
Number of pages | 8 |
Journal | Pattern Recognition Letters |
Volume | 20 |
Issue number | 1 |
DOIs | |
State | Published - 1 Jan 1999 |
Keywords
- Auto-associative neural network
- Classification
- Data projection
- Feature extraction
- Multilayer perceptron
- Principal components
- Sammon's mapping
ASJC Scopus subject areas
- Software
- Signal Processing
- Computer Vision and Pattern Recognition
- Artificial Intelligence