Evolving self-organizing maps for on-line learning, data analysis and modelling
Deng, Da; Kasabov, Nikola

View/ Open
Cite this item:
Deng, D., & Kasabov, N. (2000). Evolving self-organizing maps for on-line learning, data analysis and modelling (Information Science Discussion Papers Series No. 2000/06). University of Otago. Retrieved from http://hdl.handle.net/10523/1033
Permanent link to OUR Archive version:
http://hdl.handle.net/10523/1033
Abstract:
In real world information systems, data analysis and processing are usually needed to be done in an on-line, self-adaptive way. In this respect, neural algorithms of incremental learning and constructive network models are of increased interest. In this paper we present a new algorithm of evolving self-organizing map (ESOM), which features fast one-pass learning, dynamic network structure, and good visualisation ability. Simulations have been carried out on some benchmark data sets for classification and prediction tasks, as well as on some macroeconomic data for data analysis. Compared with other methods, ESOM achieved better classification with much shorter learning time. Its performance for time series modelling is also comparable, requiring more hidden units but with only one-pass learning. Our results demonstrate that ESOM is an effective computational model for on-line learning, data analysis and modelling.
Date:
2000-03
Publisher:
University of Otago
Pages:
16
Series number:
2000/06
Research Type:
Discussion Paper
Collections
- Knowledge Engineering Laboratory [25]
- Information Science [475]
- Discussion Paper [429]