Improving the speed and quality of extreme learning machine by conjugate gradient method
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27240%2F18%3A10241695" target="_blank" >RIV/61989100:27240/18:10241695 - isvavai.cz</a>
Result on the web
<a href="http://dx.doi.org/10.1007/978-3-319-60834-1_14" target="_blank" >http://dx.doi.org/10.1007/978-3-319-60834-1_14</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1007/978-3-319-60834-1_14" target="_blank" >10.1007/978-3-319-60834-1_14</a>
Alternative languages
Result language
angličtina
Original language name
Improving the speed and quality of extreme learning machine by conjugate gradient method
Original language description
Extreme Learning Machine (ELM) is a novel learning algorithm. It is basically a feedforward neural network with one hidden layer, fixed input weights and fixed biases. ELM has become popular in recent years due to the fast learning speed and good generalization performance. A novel approach based on Conjugate Gradient Method (CG) is proposed in this Article to improve original ELM. As experiments show, proposed approach is both faster and have higher quality on all tested datasets. Results have also shown that higher quality can be achieved after four iterations of CG. This means that expensive pseudo-inverse operation used in the original algorithm can be replaced by four matrix-vector multiplication and several scalar products. Therefore the proposed approach is more suitable for parallel architectures or can be used for larger datasets. (C) 2018, Springer International Publishing AG.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
Result continuities
Project
—
Continuities
S - Specificky vyzkum na vysokych skolach
Others
Publication year
2018
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
Advances in intelligent systems and computing. Volume 565
ISBN
978-3-319-60833-4
ISSN
2194-5357
e-ISSN
neuvedeno
Number of pages
10
Pages from-to
128-137
Publisher name
Springer
Place of publication
Berlin
Event location
Marrákeš
Event date
Nov 21, 2016
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
—