The Formula Study in Determining the Best Number of Neurons in Neural Network Backpropagation Architecture with Three Hidden Layers

Syaharuddin, Syaharuddin; Fatmawati, Fatmawati; Suprajitno, Herry

doi:10.29207/resti.v6i3.4049

Cited by 6 publications

(6 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The number of neurons in the input layer, hidden layer, and output layer was 36-73-37-19-1. This architecture is in accordance with the research results of Syaharuddin et al [53] who carried out training-testing on rainfall and temperature data from Lombok International Airport station. Furthermore, Syaharuddin et al [44,54] also tested several parameters of the backpropagation architecture with three hidden layers.…”

Section: Research Proceduressupporting

confidence: 87%

Long-Term Forecasting of Crop Water Requirement with BP-RVM Algorithm for Food Security and Harvest Risk Reduction

Syaharuddin¹,

Fatmawati²,

Suprajitno³

2023

IJSSE

View full text Add to dashboard Cite

Cropping pattern planning is important to avoid crop failure. Meanwhile, cropping patterns are affected by climate change, which is constantly shifting and erratic. Mistakes in determining the planting schedule will affect the risk of crop failure. Hence, climate forecast using long-term hydro-climatological data must be conducted as cropping patterns are mapped for a multi-year period. Data was collected from the Meteorology, Climatology, and Geophysics Agency in Lombok Island. This paper discusses the combination of backpropagation and relevance vector machine with RBF kernel. We utilized BP-RVM architecture with three hidden layers to improve the performance of the network. This combination is utilized because of the BP algorithm's ability to simplify data pattern recognition and RVM to speed up and reduce the number of iterations for each data training-testing process. The evapotranspiration of each crop was then calculated using the FAO24 Blaney-Criddle method. Based on the forecasting, the average MAPE was below 20%, which indicates "good forecasting". The evapotranspiration values of CGPRT and horticultural crops were almost the same with an average of 2.79 mm/day and 2.78 mm/day. These values are lower than the evapotranspiration values of tobacco and rice. Finally, based on the calculation of each crop's water requirement throughout the year, it was recommended to start the first planting season at the end of October. The results of this study can be recommended to the government to apply the BP-RVM algorithm in forecasting hydro-climatological data and optimizing cropping patterns to avoid crop failure and maintain the stability of national food security.

show abstract

Section: Research Proceduressupporting

confidence: 87%

Long-Term Forecasting of Crop Water Requirement with BP-RVM Algorithm for Food Security and Harvest Risk Reduction

Syaharuddin¹,

Fatmawati²,

Suprajitno³

2023

IJSSE

View full text Add to dashboard Cite

show abstract

“…Thus, from these two architectural models, it was found that the three hidden layers (100-50-10-1) have predictive results derived from the actual data and can be used for forecasting calculations in future periods; or the MSE and RMSE value of three hidden layers is smaller than that of two hidden layers. This finding is supported by the research of Syaharuddin et al [32] who argued that the determination of the best number of neurons on backpropagation using three layers with an architecture of 36-73-37-19-1 showed the highest level of accuracy at the time of precipitation data, obtaining an MSE of 0.0291 and an accuracy rate of 99.94% for data training and 99.99% for data testing.…”

Section: Training Testing and Data Prediction Phasesupporting

confidence: 58%

Study of Climate Change in the Mandalika International Circuit Area Using Neural Network Backpropagation

Mandailina¹,

Nurhalimah²,

Mehmood³

et al. 2022

RIA

View full text Add to dashboard Cite

Climate change is a global phenomenon that also causes small-scale effects. This study aims to determine future climate changes in the Mandalika International Circuit area using the artificial neural network of the MATLAB GUI-based backpropagation method. The simulation stage used daily rainfall intensity data collected in the Mandalika International Circuit area from 2012-2021 (365 data). A preliminary analysis concluded that the Mandalika International Circuit area is dominated by a very wet climate according to the Schmidt-Ferguson classification, which occurred in 2012, 2013, 2017, and 2021. This study used two architectural models with two and three hidden layers. The TRAINRP training function and the LOGSIG activation function were utilized at each hidden layer. Between the two architectures, the better architecture was selected, namely the 100-50-10-1 (three hidden layers) that resulted in an accuracy rate of 99.90% and an MSE of 0.0412376 achieved in the 258th iteration. These results indicate that the area has a very wet climate with the highest rain intensity in March and the lowest in January. The results of this study show that the backpropagation method can be used to help formulate an alternative policy on the measures for handling and mitigating extreme climate change in upcoming periods, especially during international events at the Mandalika International Circuit area.

show abstract

“…A BP neural network with three hidden layers is chosen for data training. The number of neurons in the three hidden layers is set as 17, 10, and 6 (the number of neurons in the hidden layers can be determined based on empirical formulas [52,53]). The neural network's input layer includes the eight design parameters of the actuator.…”

Section: Network Prediction Model and Its Application For Rigidly-con...mentioning

confidence: 99%

Modeling of rigidly-constrained pure shear dielectric elastomer actuators: electromechanical coupling network method

Sun,

Liang,

Zhang

et al. 2024

Smart Mater. Struct.

View full text Add to dashboard Cite

The rigidly-constrained pure shear dielectric elastomer actuator (PS-DEA) has become one of the critical configurations in linear soft actuator design due to its excellent uni-directional actuation performance and convenient preparation process. However, the theoretical analyses are primarily conducted by employing ideal models and lack consideration of the lateral necking deformation of PS-DEA, which has an essential impact on the performance evaluation and optimal design of PS-DEA. Therefore, in this paper, a user subroutine that describing the behavior of the electromechanical behavior of DE in terms of the Gent free-energy model is developed, and then a parametric model of the PS-DEA is established. Different combinations of actuator parameters are obtained by Latin hypercube sampling, and the actuator's performance under the parameters is simulated by the finite element method. The finite element results are taken as a sample set, and a BP neural network with three hidden layers is employed to train the samples and obtain a PS-DEA network prediction model, which is experimentally analyzed to validate its accuracy and effectiveness. The prediction model explores the influence of geometric and pre-stretching parameters on the actuator’s performance and obtains the difference between the ideal theoretical and the network prediction model under various parameters. The method in this paper provides a new design methodology and theoretical basis for developing high-performance dielectric elastomer actuators.

show abstract

The Formula Study in Determining the Best Number of Neurons in Neural Network Backpropagation Architecture with Three Hidden Layers

Cited by 6 publications

References 16 publications

Long-Term Forecasting of Crop Water Requirement with BP-RVM Algorithm for Food Security and Harvest Risk Reduction

Long-Term Forecasting of Crop Water Requirement with BP-RVM Algorithm for Food Security and Harvest Risk Reduction

Study of Climate Change in the Mandalika International Circuit Area Using Neural Network Backpropagation

Modeling of rigidly-constrained pure shear dielectric elastomer actuators: electromechanical coupling network method

Contact Info

Product

Resources

About