A New Adaptive Strategy for Pruning and Adding Hidden Neurons during Training Artificial Neural Networks

2008 ◽  
pp. 40-48
Author(s):  
Md. Monirul Islam ◽  
Md. Abdus Sattar ◽  
Md. Faijul Amin ◽  
Kazuyuki Murase
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Adaptive Strategy ◽  
Artificial Neural ◽  
Hidden Neurons

Towards an Improved Ensemble Learning Model of Artificial Neural Networks

2016 ◽  
pp. 762-793
Author(s):  
Fatai Anifowose ◽  
Jane Labadin ◽  
Abdulazeez Abdulraheem
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Reservoir Characterization ◽  
Evaluation Criteria ◽  
Search Space ◽  
Combination Rules ◽  
Search Spaces ◽  
Ensemble Machine Learning ◽  
Artificial Neural ◽  
Hidden Neurons

Artificial Neural Networks (ANN) have been widely applied in petroleum reservoir characterization. Despite their wide use, they are very unstable in terms of performance. Ensemble machine learning is capable of improving the performance of such unstable techniques. One of the challenges of using ANN is choosing the appropriate number of hidden neurons. Previous studies have proposed ANN ensemble models with a maximum of 50 hidden neurons in the search space thereby leaving rooms for further improvement. This chapter presents extended versions of those studies with increased search spaces using a linear search and randomized assignment of the number of hidden neurons. Using standard model evaluation criteria and novel ensemble combination rules, the results of this study suggest that having a large number of “unbiased” randomized guesses of the number of hidden neurons beyond 50 performs better than very few occurrences of those that were optimally determined.

Multilayer Backpropagation Neural Networks for Implementation of Logic Gates

2021 ◽  
Vol 12 (1) ◽  
pp. 1-12
Author(s):  
Santosh Giri ◽  
Basanta Joshi
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Network ◽  
Artificial Neural Networks ◽  
Specific Problem ◽  
Logic Gates ◽  
Backpropagation Algorithm ◽  
Processing Elements ◽  
Artificial Neural ◽  
Hidden Neurons

ANN is a computational model that is composed of several processing elements (neurons) that tries to solve a specific problem. Like the human brain, it provides the ability to learn from experiences without being explicitly programmed. This article is based on the implementation of artificial neural networks for logic gates. At first, the 3 layers Artificial Neural Network is designed with 2 input neurons, 2 hidden neurons & 1 output neuron. after that model is trained by using a backpropagation algorithm until the model satisfies the predefined error criteria (e) which set 0.01 in this experiment. The learning rate (α) used for this experiment was 0.01. The NN model produces correct output at iteration (p)= 20000 for AND, NAND & NOR gate. For OR & XOR the correct output is predicted at iteration (p)=15000 & 80000 respectively.

Towards an Improved Ensemble Learning Model of Artificial Neural Networks

2014 ◽  
pp. 76-106
Author(s):  
Fatai Anifowose ◽  
Jane Labadin ◽  
Abdulazeez Abdulraheem
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Reservoir Characterization ◽  
Evaluation Criteria ◽  
Search Space ◽  
Combination Rules ◽  
Search Spaces ◽  
Ensemble Machine Learning ◽  
Artificial Neural ◽  
Hidden Neurons

Artificial Neural Networks (ANN) have been widely applied in petroleum reservoir characterization. Despite their wide use, they are very unstable in terms of performance. Ensemble machine learning is capable of improving the performance of such unstable techniques. One of the challenges of using ANN is choosing the appropriate number of hidden neurons. Previous studies have proposed ANN ensemble models with a maximum of 50 hidden neurons in the search space thereby leaving rooms for further improvement. This chapter presents extended versions of those studies with increased search spaces using a linear search and randomized assignment of the number of hidden neurons. Using standard model evaluation criteria and novel ensemble combination rules, the results of this study suggest that having a large number of “unbiased” randomized guesses of the number of hidden neurons beyond 50 performs better than very few occurrences of those that were optimally determined.

Influence of Number of Gene Loci on Prediction of Animal Phenotype Value Using Back-Propagation Artificial Neural Network

2011 ◽  
Vol 460-461 ◽  
pp. 329-334
Author(s):  
Xue Bin Li ◽  
Xiao Ling Yu ◽  
Xiao Jian Zhang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Back Propagation ◽  
Mathematical Function ◽  
Genomic Breeding ◽  
Breeding Values ◽  
Gene Number ◽  
Artificial Neural ◽  
Hidden Neurons

Vast amount of bioinformation immerged in the past, HapMap Project had genotyped more than 3.1 million Single Nucleotide Polymorphisms (SNPs) information by 2007, a prediction equation based on SNPs was derived to calculate genomic breeding values. However, the simple mathematical function could not reflect the complex relation between genome and phenotypes. Unlike the methods of regression, artificial neural networks could perform well for optimization in complex non-linear systems; artificial neural networks have not been used to calculate genomic breeding values. In this paper, back-propagation neural network is used to simulate and predict the genomic breeding values or polygenic genotype value, and the different numbers of gene loci and hidden neurons were used to discuss the influence of the learning rate on estimating the polygenic genotype value. The result showed normalization was very important for training prediction model. After phenotype value normalized, optimum neural network for estimating the animal phenotype could be established without considering the gene number, but the optimum neural network should be selected from amount of neuronal networks with different hidden neuron number. No matter what the gene number is, as well as the number of hidden neurons is right, BP networks could be used to predict the animal phenotypes.

Human Immunoglobulin G Adsorption in Epoxy Chitosan/Alginate Adsorbents: Evaluation of Isotherms by Artificial Neural Networks

2019 ◽  
Vol 14 (4) ◽  
Author(s):  
Ana Carolina Moreno Pássaro ◽  
Tainá Maia Mozetic ◽  
Jones Erni Schmitz ◽  
Ivanildo José da Silva ◽  
Tiago Dias Martins ◽  
...  
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Adsorption Capacity ◽  
Protein Interactions ◽  
Nonlinear Models ◽  
Correlation Coefficients ◽  
Sodium Phosphate Buffer ◽  
Artificial Neural ◽  
Hidden Layer ◽  
Hidden Neurons

Abstract This work aimed to evaluate the interaction of human IgG in non-conventional adsorbents based on chitosan and alginate in the absence and presence of Reactive Green, Reactive Blue and Cibacron Blue immobilized as ligands. The adsorption was evaluated at 277, 288, 298 and 310 K using sodium phosphate buffer, pH 7.6, at 25 mmol L−1. The highest adsorption capacity was observed in the experiments performed with no immobilized dye, although all showed adsorption capacity higher than 120 mg g−1. Data modeling was done using Langmuir, Langmuir-Freundlich and Temkin classical nonlinear models, and artificial neural networks (ANN) for comparison. According to the parameters obtained, a possible adsorption in multilayers was observed due to protein-adsorbent and protein-protein interactions, concluding that IgG adsorption process is favorable and spontaneous. Using an ANN structure with 3 hidden neurons (single hidden layer), the MSE (RMSE) for training, test and validation were 13.698 (3.701), 11.206 (3.347) and 7.632 (2.763), respectively, achieving correlation coefficients of 0.999 in all steps. ANN modeling proved to be effective in predicting the adsorption isotherms in addition to overcoming the difficulties caused by experimental errors and/or arising from adsorption phenomenology.

scholarly journals Long-Lead-Time Prediction of Storm Surge Using Artificial Neural Networks and Effective Typhoon Parameters: Revisit and Deeper Insight

Water ◽  
2020 ◽  
Vol 12 (9) ◽  
pp. 2394
Author(s):  
Wei-Ting Chao ◽  
Chih-Chieh Young ◽  
Tai-Wen Hsu ◽  
Wen-Cheng Liu ◽  
Chian-Yi Liu
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Storm Surge ◽  
Lead Time ◽  
General Pattern ◽  
Disaster Mitigation ◽  
Ann Model ◽  
Artificial Brain ◽  
Artificial Neural ◽  
Hidden Neurons

Storm surge induced by severe typhoons has caused many catastrophic tragedies to coastal communities over past decades. Accurate and efficient prediction/assessment of storm surge is still an important task in order to achieve coastal disaster mitigation especially under the influence of climate change. This study revisits storm surge predictions using artificial neural networks (ANN) and effective typhoon parameters. Recent progress of storm surge modeling and some remaining unresolved issues are reviewed. In this paper, we chose the northeastern region of Taiwan as the study area, where the largest storm surge record (over 1.8 m) has been observed. To develop the ANN-based storm surge model for various lead-times (from 1 to 12 h), typhoon parameters are carefully examined and selected by analogy with the physical modeling approach. A knowledge extraction method (KEM) with backward tracking and forward exploration procedures is also proposed to analyze the roles of hidden neurons and typhoon parameters in storm surge prediction, as well as to reveal the abundant, useful information covered in the fully-trained artificial brain. Finally, the capability of ANN model for long-lead-time predictions and influences in controlling parameters are investigated. Overall, excellent agreement with observations (i.e., the coefficient of efficiency CE > 0.95 for training and CE > 0.90 for validation) is achieved in one-hour-ahead prediction. When the typhoon affects coastal waters, contributions of wind speed, central pressure deficit, and relative angle are clarified via influential hidden neurons. A general pattern of maximum storm surge under various scenarios is also obtained. Moreover, satisfactory accuracy is successfully extended to a much longer lead time (i.e., CE > 0.85 for training and CE > 0.75 for validation in 12-h-ahead prediction). Possible reasons for further accuracy improvement compared to earlier works are addressed.

scholarly journals The use of artificial neural networks for mathematical modeling of the effect of composition and production conditions on the properties of PVC floor coverings

2017 ◽  
Vol 71 (1) ◽  
pp. 11-18
Author(s):  
Rajko Radovanovic ◽  
Mirjana Jovicic ◽  
Oskar Bera ◽  
Jelena Pavlicevic ◽  
Branka Pilic ◽  
...  
Keyword(s):  
Mechanical Properties ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Processing Conditions ◽  
Complex Composition ◽  
Strongly Connected ◽  
Artificial Neural ◽  
Floor Coverings ◽  
End Use ◽  
Hidden Neurons

The application of PVC floor coverings is strongly connected with their end-use properties, which depend on the composition and processing conditions. It is very difficult to estimate the proper influence of the production parameters on the characteristics of PVC floor coverings due to their complex composition and various preparation procedures. The effect of different processing variables (such as time of bowling, temperature of bowling and composition of PVC plastisol) on the mechanical properties of PVC floor coverings was investigated. The influence of different input parameters on the mechanical properties was successfully determined using an artificial neural network with an optimized number of hidden neurons. The Garson and Yoon models were applied to calculate and describe the variable contributions in the artificial neural networks.

scholarly journals Artificial Neural Networks pruning approach for geodetic velocity field determination

2013 ◽  
Vol 19 (4) ◽  
pp. 558-573 ◽  
Author(s):  
Mustafa Yilmaz
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Velocity Field ◽  
Back Propagation ◽  
Geodetic Network ◽  
Training Data ◽  
Validation Data ◽  
Artificial Neural ◽  
Point Velocity ◽  
Hidden Neurons

There has been a need for geodetic network densification since the early days of traditional surveying. In order to densify geodetic networks in a way that will produce the most effective reference frame improvements, the crustal velocity field must be modelled. Artificial Neural Networks (ANNs) are widely used as function approximators in diverse fields of geoinformatics including velocity field determination. Deciding the number of hidden neurons required for the implementation of an arbitrary function is one of the major problems of ANN that still deserves further exploration. Generally, the number of hidden neurons is decided on the basis of experience. This paper attempts to quantify the significance of pruning away hidden neurons in ANN architecture for velocity field determination. An initial back propagation artificial neural network (BPANN) with 30 hidden neurons is educated by training data and resultant BPANN is applied on test and validation data. The number of hidden neurons is subsequently decreased, in pairs from 30 to 2, to achieve the best predicting model. These pruned BPANNs are retrained and applied on the test and validation data. Some existing methods for selecting the number of hidden neurons are also used. The results are evaluated in terms of the root mean square error (RMSE) over a study area for optimizing the number of hidden neurons in estimating densification point velocity by BPANN.

Sign in / Sign up

Export Citation Format

Share Document