MetaCartSign in to MyCiteSeer

Include Citations | Advanced Search | Help

Include Citations | Advanced Search | Help

  An incremental learning algorithm that optimizes network size and sample size in one trial (1994) [6 citations — 0 self]

Download:
Download as a PDF | Download as a PS
by Byoung-tak Zhang
In Proceedings of the IEEE International Conference on Neural Networks
ftp://www.ais.fraunhofer.de/pub/as/ga/gmd_as_ga-94_05.ps
Add To MetaCart

Abstract:

Abstract--- A constructive learning algorithm is described that builds a feedforward neural network with an optimal number of hidden units to balance convergence and generalization. The method starts with a small training set and a small network, and expands the training set incrementally after training. If the training does not converge, the network grows incrementally to increase its learning capacity. This process, called selective learning with flexible neural architectures (self), results in a construction of an optimal size network for learning all the given data using only a minimal subset of them. We show that the network size optimization combined with active example selection generalizes significantly better and converges faster than conventional methods. I.

Citations

2140 Learning Internal Representations by Error Propagation – Rumelhart, Hinton, et al. - 1986
800 Multilayer feedforward networks are universal approximators – Hornik, Stinchcombe, et al. - 1989
551 The cascade-correlation learning architecture – Fahlman, Lebiere - 1991
524 Networks for approximation and learning – Poggio, Girosi - 1990
489 Neural networks and the bias/variance dilemma – Geman, Bienenstock, et al. - 1992
293 What size net gives valid generalization – Baum, Haussler - 1989
190 The Condensed Nearest Neighbor Rule – Hart - 1968
102 Dynamic node creation in backpropagation networks – Ash - 1989
97 First and SecondOrder Methods for Learning Between Steepest Descentand Newton's Method – Battiti - 1992
75 Generalization and network design strategies – LeCun - 1989
69 Back-propagation algorithm which varies the number of hidden units – Hirose, Yamashita, et al. - 1991
48 Principles of risk minimization for learning theory – Vapnik - 1992
44 Selecting concise training sets from clean data – Plutowski, White - 1993
36 The Vapnik-Chervonenkis dimension: Information versus complexity in learning – Abu-Mostafa - 1989
26 Accelerated learning by active example selection – Zhang - 1994
21 Four types of lea.rning curves – Amari, Fujita, et al. - 1992
17 Neural networks that teach themselves through genetic discovery of novel examples – Zhang, Veenker - 1991
7 Learning by Genetic Neural Evolution – Zhang - 1992
7 Focused incremental learning for improved generalization with reduced training sets – Zhang, Veenker - 1991
6 Scaling and generalization in neural networks: a case study – Ahmad, Tesauro - 1988
5 Learning and using specific instances – Volper - 1987
4 Self-development Learning: Constructing optimal size neural networks via incremental data selection – Zhang - 1993
3 Generalization in connectionist networks that realize Boolean functions – Huyser, Horowitz - 1989
3 Neural network constructive algorithms: Trading generalization for learning efficiency – unknown authors - 1993
2 On the approximate realization of continuous mappingsby neural networks – Funahashi - 1989
1 et al., "Large automatic learning, rule extraction, and generalization – Denker - 1987
1 Hwang et al., "Query-based learning applied to partially trained multilayer perceptrons – N - 1991
1 Constructive learning by specialization – Refenes, Vithlani - 1991
View or Download | Add to My Collection | Correct Errors