Proceedings Article10.1109/ICMLC.2006.259140
A New Fast Learning Algorithm for Multi-Layer Feedforward Neural Networks
Dexian Zhang,Can Liu,Zi-Qiang Wang,Nan-bo Liu +3 more
- 01 Aug 2006
- pp 2928-2934
3
TL;DR: A new approach to accelerate learning efficiency is proposed, including linearization technique of the non-linear relation, the convergence technique based on the local equalization of training sample's errors, and the rotation adjustment of the weights.
read more
Abstract: The strong nonlinear relation between the training sample's impact on the errors and error's derivatives is the fundamental reason underlying the low learning efficiency of the multi-layer forward neural networks. Effectively decreasing the degree of the nonlinear relation and its impact on network learning is critical to improve the neural network's training efficiency. Based on the above idea, this paper propose a new approach to accelerate learning efficiency, including linearization technique of the non-linear relation, the convergence technique based on the local equalization of training sample's errors, and the rotation adjustment of the weights. A new fast learning algorithm for the multi-layer forward neural networks is also presented. The experimental results prove that the new algorithm is capable of shortening the training time by hundreds time and remarkably improve generalization of the neural networks, compared with the conventional algorithms.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
•Dissertation
Development of a Novel Equaliser for Communication Channels using Tabu search Technique in Neural Network Paradigm
K R Subhashini
- 01 Jan 2008
TL;DR: Results show that the use of TS for adapting the weights and slopes for an ANN not only improves the performance of the equalizer but also reduces the structural complexity of the ANN.
•Posted Content
A Gegenbauer Neural Network with Regularized Weights Direct Determination for Classification
Jie He,Tao Chen,Zhijun Zhang +2 more
TL;DR: A novel type neural network based on Gegenbauer orthogonal polynomials is constructed and investigated, which is a unified learning mechanism for binary and multi-class classification problems and tends to have comparable (or even better) generalization performances, computational scalability and efficiency, and classification robustness, compared to least square support vector machine.
•Dissertation
Designing an equalizer structure using gradient descent algorithms
B. Pavan Kumar
- 01 Jan 2007
TL;DR: To overcome the problem of getting trapped in local minima of the Back Propagation algorithm, a faster method of training the neural network using RLS algorithm is proposed in this thesis work.
References
Universal approximation using feedforward neural networks: a survey of some existing methods, and some new results
Franco Scarselli,Ah Chung Tsoi +1 more
TL;DR: A unifying framework is introduced to understand existing approaches to investigate the universal approximation problem using feedforward neural networks, and two training algorithms are introduced which can determine the weights of feedforward Neural Network, with sigmoidal activation neurons, to any degree of prescribed accuracy.
637
•Posted Content
NeuroRule: A Connectionist Approach to Data Mining
Hongjun Lu,Rudy Setiono,Huan Liu +2 more
TL;DR: In this article, a data mining process using neural networks with the emphasis on rule extraction is described, where rules which are similar to, or more concise than those generated by the symbolic methods can be extracted from the neural networks.
141
Unsupervised mutual information criterion for elimination of overtraining in supervised multilayer networks
TL;DR: A network architecture designed for use with a cost function that includes a novel complexity penalty term that effectively describes the network complexity with respect to the given data in an unsupervised fashion is presented.
72
Adaptive regularization parameter selection method for enhancing generalization capability of neural networks
Chi-Tat Leung,Tommy W. S. Chow +1 more
TL;DR: A novel adaptive regularization parameter selection (ARPS) method is proposed in this paper to enhance the performance of the regularization method by enabling a gradient descent type training to tunnel through some of the undesired sub-optimal solutions on the composite error surface by means of changing the value of theRegularization parameter.
A sequential learning approach for single hidden layer neural networks
Jie Zhang,A.J. Morris +1 more
TL;DR: By using mixed types of neurons, it is found that more accurate neural network models, with a smaller number of hidden neurons than seen in conventional networks, can be developed.