Speech Recognition Using Locality Preserving Projection Based on Multi Kernel Learning Supervision

DOI

10.2991/isci-15.2015.202How to use a DOI?

Keywords

Feature Extraction; Multiple-kernel learning; Supervised learning; Locality Preserving Projection; Speech recognition

Abstract

The multi kernel supervising locality preserving projective speech recognition method is proposed. It utilizes a multi-kernel learning method to project speech feature vectors to high dimensional linear space, and then uses linear mapping by locality preserving projective algorithm. It utilizes Locality Preserving Projection to lower the dimension of samples and meanwhile keep the approximate separability of different classes in feature space. Considering the different mapping vector to keep the local structure important degree is different, multiple-kernel mapping feature vectors are given different weight coefficients. Experiments show that the method has achieved dimensionality reduction, improved the retrieval speed of eigenvectors and reduced the amount of computation, and gained a good recognition effect.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Dayong Zou
AU  - Juan Wang
PY  - 2015/01
DA  - 2015/01
TI  - Speech Recognition Using Locality Preserving Projection Based on Multi Kernel Learning Supervision
BT  - Proceedings of the 2015 International Symposium on Computers & Informatics
PB  - Atlantis Press
SP  - 1508
EP  - 1516
SN  - 2352-538X
UR  - https://doi.org/10.2991/isci-15.2015.202
DO  - 10.2991/isci-15.2015.202
ID  - Zou2015/01
ER  -