Constructive feedforward neural networks using Hermite polynomial activation functions

In this paper, a constructive one-hidden-layer network is introduced where each hidden unit employs a polynomial function for its activation function that is different from other units. Specifically, both a structure level as well as a function level adaptation methodologies are utilized in construc...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transaction on neural networks and learning systems 2005-07, Vol.16 (4), p.821-833
Main Authors: Liying Ma, Khorasani, K.
Format: Article
Language:English
Subjects:
Algorithms
Applied sciences
Artificial intelligence
Backpropagation algorithms
Cluster Analysis
Computer science
control theory
systems
Computer Simulation
Computing Methodologies
Connectionism. Neural networks
Constructive neural networks
Councils
Decision Support Techniques
Exact sciences and technology
Feedforward neural networks
functional level adaptation
Hermite polynomials
Heuristic algorithms
incremental training algorithms
Models, Biological
Models, Statistical
Neural networks
Neural Networks (Computer)
Neurons
Nonhomogeneous media
Numerical Analysis, Computer-Assisted
Pattern Recognition, Automated - methods
Performance analysis
Polynomials
Stochastic Processes
Testing
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, a constructive one-hidden-layer network is introduced where each hidden unit employs a polynomial function for its activation function that is different from other units. Specifically, both a structure level as well as a function level adaptation methodologies are utilized in constructing the network. The functional level adaptation scheme ensures that the "growing" or constructive network has different activation functions for each neuron such that the network may be able to capture the underlying input-output map more effectively. The activation functions considered consist of orthonormal Hermite polynomials. It is shown through extensive simulations that the proposed network yields improved performance when compared to networks having identical sigmoidal activation functions.
ISSN:1045-9227
2162-237X
1941-0093
2162-2388
DOI:10.1109/TNN.2005.851786