Analysis of Feature Extraction Criterion Function Maximum in Nonlinear Multi-layer Feedforward Neural Networks for Pattern Recognition

Junhong, Cao; Wei, Zhehao; Huang, Tao; Xiong, Xianwei

doi:10.1109/icicta.2010.193

Cited by 2 publications

(5 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to the results shown in Fig ( 6) and Fig. (7), the best performance of GNN could be achieved in a wider range of combinations of (γ,L). Especially, the performance of GNN is not sensitive to L, and can still perform well even when there only exists a very few neurons in the network.…”

Section: E Sensitivity To Hyper Parametersmentioning

confidence: 85%

“…2) Performance when L is small: From the Fig ( 6) and Fig. (7), we can also see that when neurons are few, the generalization performance of GNN are still good, while the generalization performance of ELM grows slowly when L grows. Thus, experiments on the same two data sets are conducted to compare the generalization performance of these two algorithms when L is small.…”

Section: E Sensitivity To Hyper Parametersmentioning

confidence: 88%

“…Proof: Without loss of generality, in light of Eq. ( 6) and (7), b n,kj (x j ) can be approximated by GPS as follows:…”

Section: B Gps For Multivariate Function Approximationmentioning

confidence: 99%

“…For the development of this algorithm, many researchers have explored the abilities of multi-layer FNNs [5]- [7]. Leshno has proved that multi-layer FNNs with activation functions are able to approximate any function, including continuous functions and non-sequence functions [8].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Gegenbauer Neural Network with Regularized Weights Direct Determination for Classification

He,

Chen,

Zhang

2019

Preprint

View full text Add to dashboard Cite

Single-hidden layer feed forward neural networks (SLFNs) are widely used in pattern classification problems, but a huge bottleneck encountered is the slow speed and poor performance of the traditional iterative gradient-based learning algorithms. Although the famous extreme learning machine (ELM) has successfully addressed the problems of slow convergence, it still has computational robustness problems brought by input weights and biases randomly assigned. Thus, in order to overcome the aforementioned problems, in this paper, a novel type neural network based on Gegenbauer orthogonal polynomials, termed as GNN, is constructed and investigated. This model could overcome the computational robustness problems of ELM, while still has comparable structural simplicity and approximation capability. Based on this, we propose a regularized weights direct determination (R-WDD) based on equality-constrained optimization to determine the optimal output weights. The R-WDD tends to minimize the empirical risks and structural risks of the network, thus to lower the risk of over fitting and improve the generalization ability. This leads us to a the final GNN with R-WDD, which is a unified learning mechanism for binary and multi-class classification problems. Finally, as is verified in the various comparison experiments, GNN with R-WDD tends to have comparable (or even better) generalization performances, computational scalability and efficiency, and classification robustness, compared to least square support vector machine (LS-SVM), ELM with Gaussian kernel.

show abstract

Section: E Sensitivity To Hyper Parametersmentioning

confidence: 85%

Section: E Sensitivity To Hyper Parametersmentioning

confidence: 88%

“…Proof: Without loss of generality, in light of Eq. ( 6) and (7), b n,kj (x j ) can be approximated by GPS as follows:…”

Section: B Gps For Multivariate Function Approximationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Gegenbauer Neural Network with Regularized Weights Direct Determination for Classification

He,

Chen,

Zhang

2019

Preprint

View full text Add to dashboard Cite

show abstract

“…To answer the first challenge, in terms of identifying the types of NN modelers that provide consistent performance with our system, a collection of well-known NNs and their cascaded versions are considered. These NN modelers include Single-Layer FeedForward Neural Network (SLNN), 32 Multilayer FeedForward Neural Network (MLNN), 33 SLNN-cascade, MLNN-cascade, radial-based NN (RBNN), 34 RBNN-cascade, generalized regression NN (GRNN), 35 exact radial basis network (ERBNN), 36 ERBNN-cascade, and Cascade-forward NN. A detailed comparison between the performances of these NNs is presented in the Supplementary Data .…”

Section: Analysis and Testing Of Curvaturementioning

confidence: 99%

Novel Design of a Soft Lightweight Pneumatic Continuum Robot Arm with Decoupled Variable Stiffness and Positioning

et al. 2018

View full text Add to dashboard Cite

Soft robot arms possess unique capabilities when it comes to adaptability, flexibility, and dexterity. In addition, soft systems that are pneumatically actuated can claim high power-to-weight ratio. One of the main drawbacks of pneumatically actuated soft arms is that their stiffness cannot be varied independently from their end-effector position in space. The novel robot arm physical design presented in this article successfully decouples its end-effector positioning from its stiffness. An experimental characterization of this ability is coupled with a mathematical analysis. The arm combines the light weight, high payload to weight ratio and robustness of pneumatic actuation with the adaptability and versatility of variable stiffness. Light weight is a vital component of the inherent safety approach to physical human-robot interaction. To characterize the arm, a neural network analysis of the curvature of the arm for different input pressures is performed. The curvature-pressure relationship is also characterized experimentally.

show abstract

Analysis of Feature Extraction Criterion Function Maximum in Nonlinear Multi-layer Feedforward Neural Networks for Pattern Recognition

Cited by 2 publications

References 5 publications

A Gegenbauer Neural Network with Regularized Weights Direct Determination for Classification

A Gegenbauer Neural Network with Regularized Weights Direct Determination for Classification

Novel Design of a Soft Lightweight Pneumatic Continuum Robot Arm with Decoupled Variable Stiffness and Positioning

Contact Info

Product

Resources

About