Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A novel neural network for data mining

 
: Chan, T.K.; Tan, E.C.; Haralalka, N.

Fukushima, K. ; Chinese Neural Network Council -CNNC-:
8th International Conference on Neural Information Processing 2001. Proceedings
Shanghai: Fudan University Press, 2001
pp.559-563 (Vol.2)
International Conference on Neural Information Processing (ICONIP) <8, 2001, Shanghai>
English
Conference Paper
Fraunhofer IGD ()
data mining

Abstract
The algorithm proposed here for data mining deals with the standard Multilayer Perceptron using Temporal Backpropagation algorithm with the concept of Bollinger Band Crossover from the concept of Trading Systems added to it, and is referred to as the Bollinger Band Crossover Supervised Network (BBCSN). To visualize the performances of each of this algorithm, a portfolio management scheme was designed and translated into code using Visual C++. This enables a critical analysis of the algorithm, with respect to their financial performances. According to the results obtained in this project, the newly proposed and designed Bollinger Band Crossover gives much better results than generally obtained from a Multilayer Perceptron network.

: http://publica.fraunhofer.de/documents/N-15476.html