Improving neural networks classification through chaining

Authors:
Khobaib Zaamout;John Z. Zhang
Affiliations:
Department of Mathematics and Computer Science, University of Lethbridge, Lethbridge, AB, Canada;Department of Mathematics and Computer Science, University of Lethbridge, Lethbridge, AB, Canada
Venue:
ICANN'12 Proceedings of the 22nd international conference on Artificial Neural Networks and Machine Learning - Volume Part II
Year:
2012

Citing 10
Cited 0

The upstart algorithm: a method for constructing and training feedforward neural networks

Neural Computation
The cascade-correlation learning architecture

Advances in neural information processing systems 2
Optimal brain damage

Advances in neural information processing systems 2
Neural Network Ensembles

IEEE Transactions on Pattern Analysis and Machine Intelligence
Correlation-based Feature Selection for Discrete and Numeric Class Machine Learning

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
CropAssist, an automated system for direct measurement of greenhouse tomato growth and water use

Computers and Electronics in Agriculture
The feature selection problem: traditional methods and a new algorithm

AAAI'92 Proceedings of the tenth national conference on Artificial intelligence
An empirical evaluation of bagging and boosting

AAAI'97/IAAI'97 Proceedings of the fourteenth national conference on artificial intelligence and ninth conference on Innovative applications of artificial intelligence
Neural networks for classification: a survey

IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews
Task decomposition and module combination based on class relations: a modular neural network for pattern classification

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

We present a new ensemble technique, namely chaining neural networks, as our efforts to improve neural classification. We show that using predictions of a neural network as input to another neural network trained on the same dataset will improve classification. We propose two variations of this approach, single-link and multi-link chaining. Both variations include predictions of trained neural networks in the construction and training of a new network and then store them for later predictions. In this initial work, the effectiveness of our proposed approach is demonstrated through a series of experiments on real and synthetic datasets.