Boundedness and Convergence of Online Gradient Method With Penalty for Feedforward Neural Networks

Zhang, Huisheng; Wu, Wei; Liu, Fei; Yao, Mingchen

doi:10.1109/tnn.2009.2020848

Cited by 66 publications

(5 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The main reason for this is that the ELM training mechanism is different from those of the BP and RBF networks. The training mechanism of the BP network is as follows [27,28]. In the BP networks, the data go through the input and hidden layers to the output layer and then the output value is obtained.…”

Section: The Discussion Of Training Timementioning

confidence: 99%

The study of using an extreme learning machine for rapid concentration estimation in multi-component gas mixtures

Zhao

Wang

et al. 2012

Meas. Sci. Technol.

View full text Add to dashboard Cite

Cross-sensitivity is one of the major unpleasant characteristics of metal oxide gas sensors. In order to solve this challenging problem, artificial neural networks have proved to be very powerful tools, among which back propagation (BP) and radial basis function (RBF) neural networks are the two most commonly used ones in data analysis of metal oxide gas sensors and arrays. However, relatively long training time is the major disadvantage for the BP and RBF neural networks. In order to solve this problem, an extreme leaning machine (ELM) is introduced and studied in this paper. Experimental results show that ELM networks can achieve 466 and 333 times faster training speed than the BP and RBF neural networks, respectively. In addition, ELM networks can achieve comparable concentration prediction accuracy to RBF networks which is much better than BP networks.

show abstract

Section: The Discussion Of Training Timementioning

confidence: 99%

The study of using an extreme learning machine for rapid concentration estimation in multi-component gas mixtures

Zhao

Wang

et al. 2012

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…They have a wide range of applications across almost all areas of science and engineering. Unlike conventional data processing methods, which require extensive expert knowledge when used in the modelling of electrical machines, neural networks provide a model-free, adaptive, fault-tolerant, parallel, and distributed processing solution [25].…”

Section: Neural Networkmentioning

confidence: 99%

“…The process begins with the receipt of inputs by the input layer neurons, which are processed by each layer sequentially to the output layer. The connections between neurons are unidirectional and neurons in the same layer are unconnected to each other, as shown in Figure 4 [25].…”

Section: Structurementioning

confidence: 99%

Design Synchronous Generator Using Taguchi-Based Multi-Objective Optimization

Cheng

Hong

et al. 2020

Energies

View full text Add to dashboard Cite

The extensive use of finite element models accurately simulates the temperature distribution of electrical machines. The simulation model can be quickly modified to reflect changes in design. However, the long runtime of the simulation prevents any direct application of the optimization algorithm. In this paper, research focused on improving efficiency with which expensive analysis (finite element method) is used in generator temperature distribution. A novel surrogate model based optimization method is presented. First, the Taguchi orthogonal array relates a series of stator geometric parameters as input and the temperatures of a generator as output by sampling the design decision space. A number of stator temperature designs were generated and analyzed using 3-D multi-physical field collaborative finite element model. A suitable shallow neural network was then selected and fitted to the available data to obtain a continuous optimization objective function. The accuracy of the function was verified using randomly generated geometric parameters to the extent that they were feasible. Finally, a multi-objective genetic optimization algorithm was applied in the function to reduce the average and maximum temperature of the machine simultaneously. As a result, when the Pareto front was compared with the initial data, these temperatures showed a significant decrease.

show abstract

“…In this section, the convergence of the proposed fractional-order BP neural network is analyzed. According to previous studies [ 39 – 42 ], there are four necessary conditions for the convergence of BP neural networks:…”

Section: Convergence Analysismentioning

confidence: 99%

Fractional-Order Deep Backpropagation Neural Network

Bao

Zhang

2018

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

In recent years, the research of artificial neural networks based on fractional calculus has attracted much attention. In this paper, we proposed a fractional-order deep backpropagation (BP) neural network model with L2 regularization. The proposed network was optimized by the fractional gradient descent method with Caputo derivative. We also illustrated the necessary conditions for the convergence of the proposed network. The influence of L2 regularization on the convergence was analyzed with the fractional-order variational method. The experiments have been performed on the MNIST dataset to demonstrate that the proposed network was deterministically convergent and can effectively avoid overfitting.

show abstract

Boundedness and Convergence of Online Gradient Method With Penalty for Feedforward Neural Networks

Cited by 66 publications

References 16 publications

The study of using an extreme learning machine for rapid concentration estimation in multi-component gas mixtures

The study of using an extreme learning machine for rapid concentration estimation in multi-component gas mixtures

Design Synchronous Generator Using Taguchi-Based Multi-Objective Optimization

Fractional-Order Deep Backpropagation Neural Network

Contact Info

Product

Resources

About