Abstract
The initialisation of a neural network implementation of Sammon's mapping, either randomly or based on the principal components (PCs) of the sample covariance matrix, is experimentally investigated. When PCs are employed, fewer experiments are needed and the network configuration can be set precisely without trial-and-error experimentation. Tested on five real-world databases, it is shown that very few PCs ate required to achieve a shorter training period, lower mapping error and higher classification accuracy, compared with those based on random initialisation.
Original language | English |
---|---|
Pages (from-to) | 61-68 |
Number of pages | 8 |
Journal | Pattern Analysis and Applications |
Volume | 3 |
Issue number | 1 |
DOIs | |
State | Published - 1 Jan 2000 |
Keywords
- Classification
- Data projection
- Initialisation
- Neural networks
- Principal component analysis (PCA)
- Sammon's mapping
ASJC Scopus subject areas
- Computer Vision and Pattern Recognition
- Artificial Intelligence