Loading...
Loading...

Go to the content (press return)

Neural networks with periodic and monotonic activation functions: a comparative study in classification problems

Author
Romero, E.; Sopena, J.; Alquezar, R.; Moliner, J.
Type of activity
Report
Date
2000-02
Code
LSI-00-12-R
Repository
http://hdl.handle.net/2117/85060 Open in new window
Abstract
This article discusses a number of reasons why the use of non-monotonic functions as activation functions can lead to a marked improvement in the performance of a neural network. Using a wide range of benchmarks we show that a multilayer feed-forward network using sine activation functions (and an appropriate choice of initial parameters) learns much faster than one incorporating sigmoid functions - as much as 150-500 times faster - when both types are trained with backpropagation. Learning spee...
Citation
Romero, E., Sopena, J., Alquézar, R., Moliner, J. "Neural networks with periodic and monotonic activation functions: a comparative study in classification problems". 2000.
Keywords
Neural networks, Non-monotonic functions
Group of research
IDEAI-UPC - Intelligent Data Science and Artificial Intelligence Research Center
SOCO - Soft Computing