Optimization of Number of Neurons in the Hidden Layer in Feed Forward Neural ...IJERA Editor
The architectures of Artificial Neural Networks (ANN) are based on the problem domain and it is applied during
the „training phase‟ of sample data and used to infer results for the remaining data in the testing phase.
Normally, the architecture consist of three layers as input, hidden, output layers with the number of nodes in the
input layer as number of known values on hand and the number of nodes as result to be computed out of the
values of input nodes and hidden nodes as the output layer. The number of nodes in the hidden layer is
heuristically decided so that the optimum value is obtained with reasonable number of iterations with other
parameters with its default values. This study mainly focuses on Cascade-Correlation Neural Networks (CCNN)
using Back-Propagation (BP) algorithm which finds the number of neurons during the training phase itself by
appending one from the previous iteration satisfying the error condition gives a promising result on the optimum
number of neurons in the hidden layer
Black-box modeling of nonlinear system using evolutionary neural NARX modelIJECEIAES
Nonlinear systems with uncertainty and disturbance are very difficult to model using mathematic approach. Therefore, a black-box modeling approach without any prior knowledge is necessary. There are some modeling approaches have been used to develop a black box model such as fuzzy logic, neural network, and evolution algorithms. In this paper, an evolutionary neural network by combining a neural network and a modified differential evolution algorithm is applied to model a nonlinear system. The feasibility and effectiveness of the proposed modeling are tested on a piezoelectric actuator SISO system and an experimental quadruple tank MIMO system.
Comparison of Neural Network Training Functions for Hematoma Classification i...IOSR Journals
Classification is one of the most important task in application areas of artificial neural networks
(ANN).Training neural networks is a complex task in the supervised learning field of research. The main
difficulty in adopting ANN is to find the most appropriate combination of learning, transfer and training
function for the classification task. We compared the performances of three types of training algorithms in feed
forward neural network for brain hematoma classification. In this work we have selected Gradient Descent
based backpropagation, Gradient Descent with momentum, Resilence backpropogation algorithms. Under
conjugate based algorithms, Scaled Conjugate back propagation, Conjugate Gradient backpropagation with
Polak-Riebreupdates(CGP) and Conjugate Gradient backpropagation with Fletcher-Reeves updates (CGF).The
last category is Quasi Newton based algorithm, under this BFGS, Levenberg-Marquardt algorithms are
selected. Proposed work compared training algorithm on the basis of mean square error, accuracy, rate of
convergence and correctness of the classification. Our conclusion about the training functions is based on the
simulation results
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
Optimization of Number of Neurons in the Hidden Layer in Feed Forward Neural ...IJERA Editor
The architectures of Artificial Neural Networks (ANN) are based on the problem domain and it is applied during
the „training phase‟ of sample data and used to infer results for the remaining data in the testing phase.
Normally, the architecture consist of three layers as input, hidden, output layers with the number of nodes in the
input layer as number of known values on hand and the number of nodes as result to be computed out of the
values of input nodes and hidden nodes as the output layer. The number of nodes in the hidden layer is
heuristically decided so that the optimum value is obtained with reasonable number of iterations with other
parameters with its default values. This study mainly focuses on Cascade-Correlation Neural Networks (CCNN)
using Back-Propagation (BP) algorithm which finds the number of neurons during the training phase itself by
appending one from the previous iteration satisfying the error condition gives a promising result on the optimum
number of neurons in the hidden layer
Black-box modeling of nonlinear system using evolutionary neural NARX modelIJECEIAES
Nonlinear systems with uncertainty and disturbance are very difficult to model using mathematic approach. Therefore, a black-box modeling approach without any prior knowledge is necessary. There are some modeling approaches have been used to develop a black box model such as fuzzy logic, neural network, and evolution algorithms. In this paper, an evolutionary neural network by combining a neural network and a modified differential evolution algorithm is applied to model a nonlinear system. The feasibility and effectiveness of the proposed modeling are tested on a piezoelectric actuator SISO system and an experimental quadruple tank MIMO system.
Comparison of Neural Network Training Functions for Hematoma Classification i...IOSR Journals
Classification is one of the most important task in application areas of artificial neural networks
(ANN).Training neural networks is a complex task in the supervised learning field of research. The main
difficulty in adopting ANN is to find the most appropriate combination of learning, transfer and training
function for the classification task. We compared the performances of three types of training algorithms in feed
forward neural network for brain hematoma classification. In this work we have selected Gradient Descent
based backpropagation, Gradient Descent with momentum, Resilence backpropogation algorithms. Under
conjugate based algorithms, Scaled Conjugate back propagation, Conjugate Gradient backpropagation with
Polak-Riebreupdates(CGP) and Conjugate Gradient backpropagation with Fletcher-Reeves updates (CGF).The
last category is Quasi Newton based algorithm, under this BFGS, Levenberg-Marquardt algorithms are
selected. Proposed work compared training algorithm on the basis of mean square error, accuracy, rate of
convergence and correctness of the classification. Our conclusion about the training functions is based on the
simulation results
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
Incorporating Kalman Filter in the Optimization of Quantum Neural Network Par...Waqas Tariq
Kalman filter have been used for the estimation of instantaneous states of linear dynamic systems. It is a good tool for inferring of missing information from noisy measurement. The quantum neural network is another approach to the merging of fuzzy logic with the neural network and that by the investment of quantum mechanics theory in building the structure of neural network. The gradient descent algorithm has been used, widely, in training the neural network, but the problem of local minima is one of the disadvantages of this algorithm. This paper presents an algorithm to train the quantum neural network by using the extended kalman filter.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
CAPSULE NETWORK PERFORMANCE WITH AUTONOMOUS NAVIGATIONijaia
Capsule Networks (CapsNets) have been proposed as an alternative to Convolutional Neural Networks (CNNs). This paper showcases how CapsNets are more capable than CNNs for autonomous agent
exploration of realistic scenarios. In real world navigation, rewards external to agents may be rare. In turn, reinforcement learning algorithms can struggle to form meaningful policy functions. This paper’s approach Capsules Exploration Module (Caps-EM) pairs a CapsNets architecture with an Advantage
Actor Critic algorithm. Other approaches for navigating sparse environments require intrinsic reward generators, such as the Intrinsic Curiosity Module (ICM) and Augmented Curiosity Modules (ACM). CapsEM uses a more compact architecture without need for intrinsic rewards. Tested using ViZDoom, the CapsEM uses 44% and 83% fewer trainable network parameters than the ICM and Depth-Augmented Curiosity
Module (D-ACM), respectively, for 1141% and 437% average time improvement over the ICM and DACM, respectively, for converging to a policy function across "My Way Home" scenarios.
Artificial Neural Networks (ANNS) For Prediction of California Bearing Ratio ...IJMER
The behaviour of soil at the location of the project and interactions of the earth materials during and after construction has a major influence on the success, economy and safety of the work. Another complexity associated with some geotechnical engineering materials, such as sand and gravel, is the difficulty in obtaining undisturbed samples and time consuming involving skilled
technician. Knowledge of California Bearing Ratio (C.B.R) is essential in finding the road thickness. To cope up with the difficulties involved, an attempt has been made to model C.B.R in terms of Fine Fraction, Liquid Limit, Plasticity Index, Maximum Dry density, and Optimum Moisture content. A multi-layer perceptron network with feed forward back propagation is used to model varying the
number of hidden layers. For this purposes 50 soils test data was collected from the laboratory test
results. Among the test data 30 soils data is used for training and remaining 20 soils for testing using
60-40 distribution. The architectures developed are 5-4-1, 5-5-1, and 5-6-1. Model with 5-6-1 architecture is found to be quite satisfactory in predicting C.B.R of soils. A graph is plotted between
the predicted values and observed values of outputs for training and testing process, from the graph it
is found that all the points are close to equality line, indicating predicted values are close to observed
values
CAPSULE NETWORK PERFORMANCE WITH AUTONOMOUS NAVIGATIONgerogepatton
Capsule Networks (CapsNets) have been proposed as an alternative to Convolutional Neural Networks (CNNs). This paper showcases how CapsNets are more capable than CNNs for autonomous agent exploration of realistic scenarios. In real world navigation, rewards external to agents may be rare. In turn, reinforcement learning algorithms can struggle to form meaningful policy functions. This paper’s approach Capsules Exploration Module (Caps-EM) pairs a CapsNets architecture with an Advantage Actor Critic algorithm. Other approaches for navigating sparse environments require intrinsic reward generators, such as the Intrinsic Curiosity Module (ICM) and Augmented Curiosity Modules (ACM). CapsEM uses a more compact architecture without need for intrinsic rewards. Tested using ViZDoom, the CapsEM uses 44% and 83% fewer trainable network parameters than the ICM and Depth-Augmented Curiosity Module (D-ACM), respectively, for 1141% and 437% average time improvement over the ICM and DACM, respectively, for converging to a policy function across "My Way Home" scenarios
Associative memory implementation with artificial neural networkseSAT Publishing House
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology
NETWORK LEARNING AND TRAINING OF A CASCADED LINK-BASED FEED FORWARD NEURAL NE...ijaia
Presently, considering the technological advancement of our modern world, we are in dire need for a system that can learn new concepts and give decisions on its own. Hence the Artificial Neural Network is all that is required in the contemporary situation. In this paper, CLBFFNN is presented as a special and intelligent form of artificial neural networks that has the capability to adapt to training and learning of new ideas and be able to give decisions in a trimodal biometric system involving fingerprints, face and iris biometric data. It gives an overview of neural networks.
Mobile Network Coverage Determination at 900MHz for Abuja Rural Areas using A...ijtsrd
This study proposes Artificial Neural Network ANN based field strength prediction models for the rural areas of Abuja, the federal capital territory of Nigeria. The ANN based models were created on bases of the Generalized Regression Neural network GRNN and the Multi Layer Perceptron Neural Network MLP NN . These networks were created, trained and tested for field strength prediction using received power data recorded at 900MHz from multiple Base Transceiver Stations BTSs distributed across the rural areas. Results indicate that the GRNN and MLP NN based models with Root Mean Squared Error RMSE values of 4.78dBm and 5.56dBm respectively, offer significant improvement over the empirical Hata Okumura counterpart, which overestimates the signal strength by an RMSE value of 20.17dBm. Deme C. Abraham ""Mobile Network Coverage Determination at 900MHz for Abuja Rural Areas using Artificial Neural Networks"" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-4 | Issue-2 , February 2020,
URL: https://www.ijtsrd.com/papers/ijtsrd30228.pdf
Paper Url : https://www.ijtsrd.com/computer-science/artificial-intelligence/30228/mobile-network-coverage-determination-at-900mhz-for-abuja-rural-areas-using-artificial-neural-networks/deme-c-abraham
Optimized Neural Network for Classification of Multispectral ImagesIDES Editor
The proposed work involves the multiobjective PSO
based optimization of artificial neural network structure for
the classification of multispectral satellite images. The neural
network is used to classify each image pixel in various land
cove types like vegetations, waterways, man-made structures
and road network. It is per pixel supervised classification using
spectral bands (original feature space). Use of neural network
for classification requires selection of most discriminative
spectral bands and determination of optimal number of nodes
in hidden layer. We propose new methodology based on
multiobjective particle swarm optimization (MOPSO) to
determine discriminative spectral bands and the number of
hidden layer node simultaneously. The result obtained using
such optimized neural network is compared with that of
traditional classifiers like MLC and Euclidean classifier. The
performance of all classifiers is evaluated quantitatively using
Xie-Beni and â indexes. The result shows the superiority of
the proposed method.
International Journal of Computational Engineering Research(IJCER) is an intentional online Journal in English monthly publishing journal. This Journal publish original research work that contributes significantly to further the scientific knowledge in engineering and Technology
Modeling of neural image compression using gradient decent technologytheijes
The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.
The papers for publication in The International Journal of Engineering& Science are selected through rigorous peer reviews to ensure originality, timeliness, relevance, and readability.
Theoretical work submitted to the Journal should be original in its motivation or modeling structure. Empirical analysis should be based on a theoretical framework and should be capable of replication. It is expected that all materials required for replication (including computer programs and data sets) should be available upon request to the authors.
The International Journal of Engineering & Science would take much care in making your article published without much delay with your kind cooperation
Dynamic K-Means Algorithm for Optimized Routing in Mobile Ad Hoc Networks IJCSES Journal
In this paper, a dynamic K-means algorithm to improve the routing process in Mobile Ad-Hoc networks
(MANETs) is presented. Mobile ad-hoc networks are a collocation of mobile wireless nodes that can
operate without using focal access points, pre-existing infrastructures, or a centralized management point.
In MANETs, the quick motion of nodes modifies the topology of network. This feature of MANETS is lead
to various problems in the routing process such as increase of the overhead massages and inefficient
routing between nodes of network. A large variety of clustering methods have been developed for
establishing an efficient routing process in MANETs. Routing is one of the crucial topics which are having
significant impact on MANETs performance. The K-means algorithm is one of the effective clustering
methods aimed to reduce routing difficulties related to bandwidth, throughput and power consumption.
This paper proposed a new K-means clustering algorithm to find out optimal path from source node to
destinations node in MANETs. The main goal of proposed approach which is called the dynamic K-means
clustering methods is to solve the limitation of basic K-means method like permanent cluster head and fixed
cluster members. The experimental results demonstrate that using dynamic K-means scheme enhance the
performance of routing process in Mobile ad-hoc networks.
International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity
Incorporating Kalman Filter in the Optimization of Quantum Neural Network Par...Waqas Tariq
Kalman filter have been used for the estimation of instantaneous states of linear dynamic systems. It is a good tool for inferring of missing information from noisy measurement. The quantum neural network is another approach to the merging of fuzzy logic with the neural network and that by the investment of quantum mechanics theory in building the structure of neural network. The gradient descent algorithm has been used, widely, in training the neural network, but the problem of local minima is one of the disadvantages of this algorithm. This paper presents an algorithm to train the quantum neural network by using the extended kalman filter.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
CAPSULE NETWORK PERFORMANCE WITH AUTONOMOUS NAVIGATIONijaia
Capsule Networks (CapsNets) have been proposed as an alternative to Convolutional Neural Networks (CNNs). This paper showcases how CapsNets are more capable than CNNs for autonomous agent
exploration of realistic scenarios. In real world navigation, rewards external to agents may be rare. In turn, reinforcement learning algorithms can struggle to form meaningful policy functions. This paper’s approach Capsules Exploration Module (Caps-EM) pairs a CapsNets architecture with an Advantage
Actor Critic algorithm. Other approaches for navigating sparse environments require intrinsic reward generators, such as the Intrinsic Curiosity Module (ICM) and Augmented Curiosity Modules (ACM). CapsEM uses a more compact architecture without need for intrinsic rewards. Tested using ViZDoom, the CapsEM uses 44% and 83% fewer trainable network parameters than the ICM and Depth-Augmented Curiosity
Module (D-ACM), respectively, for 1141% and 437% average time improvement over the ICM and DACM, respectively, for converging to a policy function across "My Way Home" scenarios.
Artificial Neural Networks (ANNS) For Prediction of California Bearing Ratio ...IJMER
The behaviour of soil at the location of the project and interactions of the earth materials during and after construction has a major influence on the success, economy and safety of the work. Another complexity associated with some geotechnical engineering materials, such as sand and gravel, is the difficulty in obtaining undisturbed samples and time consuming involving skilled
technician. Knowledge of California Bearing Ratio (C.B.R) is essential in finding the road thickness. To cope up with the difficulties involved, an attempt has been made to model C.B.R in terms of Fine Fraction, Liquid Limit, Plasticity Index, Maximum Dry density, and Optimum Moisture content. A multi-layer perceptron network with feed forward back propagation is used to model varying the
number of hidden layers. For this purposes 50 soils test data was collected from the laboratory test
results. Among the test data 30 soils data is used for training and remaining 20 soils for testing using
60-40 distribution. The architectures developed are 5-4-1, 5-5-1, and 5-6-1. Model with 5-6-1 architecture is found to be quite satisfactory in predicting C.B.R of soils. A graph is plotted between
the predicted values and observed values of outputs for training and testing process, from the graph it
is found that all the points are close to equality line, indicating predicted values are close to observed
values
CAPSULE NETWORK PERFORMANCE WITH AUTONOMOUS NAVIGATIONgerogepatton
Capsule Networks (CapsNets) have been proposed as an alternative to Convolutional Neural Networks (CNNs). This paper showcases how CapsNets are more capable than CNNs for autonomous agent exploration of realistic scenarios. In real world navigation, rewards external to agents may be rare. In turn, reinforcement learning algorithms can struggle to form meaningful policy functions. This paper’s approach Capsules Exploration Module (Caps-EM) pairs a CapsNets architecture with an Advantage Actor Critic algorithm. Other approaches for navigating sparse environments require intrinsic reward generators, such as the Intrinsic Curiosity Module (ICM) and Augmented Curiosity Modules (ACM). CapsEM uses a more compact architecture without need for intrinsic rewards. Tested using ViZDoom, the CapsEM uses 44% and 83% fewer trainable network parameters than the ICM and Depth-Augmented Curiosity Module (D-ACM), respectively, for 1141% and 437% average time improvement over the ICM and DACM, respectively, for converging to a policy function across "My Way Home" scenarios
Associative memory implementation with artificial neural networkseSAT Publishing House
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology
NETWORK LEARNING AND TRAINING OF A CASCADED LINK-BASED FEED FORWARD NEURAL NE...ijaia
Presently, considering the technological advancement of our modern world, we are in dire need for a system that can learn new concepts and give decisions on its own. Hence the Artificial Neural Network is all that is required in the contemporary situation. In this paper, CLBFFNN is presented as a special and intelligent form of artificial neural networks that has the capability to adapt to training and learning of new ideas and be able to give decisions in a trimodal biometric system involving fingerprints, face and iris biometric data. It gives an overview of neural networks.
Mobile Network Coverage Determination at 900MHz for Abuja Rural Areas using A...ijtsrd
This study proposes Artificial Neural Network ANN based field strength prediction models for the rural areas of Abuja, the federal capital territory of Nigeria. The ANN based models were created on bases of the Generalized Regression Neural network GRNN and the Multi Layer Perceptron Neural Network MLP NN . These networks were created, trained and tested for field strength prediction using received power data recorded at 900MHz from multiple Base Transceiver Stations BTSs distributed across the rural areas. Results indicate that the GRNN and MLP NN based models with Root Mean Squared Error RMSE values of 4.78dBm and 5.56dBm respectively, offer significant improvement over the empirical Hata Okumura counterpart, which overestimates the signal strength by an RMSE value of 20.17dBm. Deme C. Abraham ""Mobile Network Coverage Determination at 900MHz for Abuja Rural Areas using Artificial Neural Networks"" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-4 | Issue-2 , February 2020,
URL: https://www.ijtsrd.com/papers/ijtsrd30228.pdf
Paper Url : https://www.ijtsrd.com/computer-science/artificial-intelligence/30228/mobile-network-coverage-determination-at-900mhz-for-abuja-rural-areas-using-artificial-neural-networks/deme-c-abraham
Optimized Neural Network for Classification of Multispectral ImagesIDES Editor
The proposed work involves the multiobjective PSO
based optimization of artificial neural network structure for
the classification of multispectral satellite images. The neural
network is used to classify each image pixel in various land
cove types like vegetations, waterways, man-made structures
and road network. It is per pixel supervised classification using
spectral bands (original feature space). Use of neural network
for classification requires selection of most discriminative
spectral bands and determination of optimal number of nodes
in hidden layer. We propose new methodology based on
multiobjective particle swarm optimization (MOPSO) to
determine discriminative spectral bands and the number of
hidden layer node simultaneously. The result obtained using
such optimized neural network is compared with that of
traditional classifiers like MLC and Euclidean classifier. The
performance of all classifiers is evaluated quantitatively using
Xie-Beni and â indexes. The result shows the superiority of
the proposed method.
International Journal of Computational Engineering Research(IJCER) is an intentional online Journal in English monthly publishing journal. This Journal publish original research work that contributes significantly to further the scientific knowledge in engineering and Technology
Modeling of neural image compression using gradient decent technologytheijes
The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.
The papers for publication in The International Journal of Engineering& Science are selected through rigorous peer reviews to ensure originality, timeliness, relevance, and readability.
Theoretical work submitted to the Journal should be original in its motivation or modeling structure. Empirical analysis should be based on a theoretical framework and should be capable of replication. It is expected that all materials required for replication (including computer programs and data sets) should be available upon request to the authors.
The International Journal of Engineering & Science would take much care in making your article published without much delay with your kind cooperation
Dynamic K-Means Algorithm for Optimized Routing in Mobile Ad Hoc Networks IJCSES Journal
In this paper, a dynamic K-means algorithm to improve the routing process in Mobile Ad-Hoc networks
(MANETs) is presented. Mobile ad-hoc networks are a collocation of mobile wireless nodes that can
operate without using focal access points, pre-existing infrastructures, or a centralized management point.
In MANETs, the quick motion of nodes modifies the topology of network. This feature of MANETS is lead
to various problems in the routing process such as increase of the overhead massages and inefficient
routing between nodes of network. A large variety of clustering methods have been developed for
establishing an efficient routing process in MANETs. Routing is one of the crucial topics which are having
significant impact on MANETs performance. The K-means algorithm is one of the effective clustering
methods aimed to reduce routing difficulties related to bandwidth, throughput and power consumption.
This paper proposed a new K-means clustering algorithm to find out optimal path from source node to
destinations node in MANETs. The main goal of proposed approach which is called the dynamic K-means
clustering methods is to solve the limitation of basic K-means method like permanent cluster head and fixed
cluster members. The experimental results demonstrate that using dynamic K-means scheme enhance the
performance of routing process in Mobile ad-hoc networks.
International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity
A Study on Fire Detection System using Statistic Color ModelIOSR Journals
Abstract: Normally fire detection system uses the heuristic fixed threshold values in their specific methods. However, input images may be changed, in general, so the heuristic fixed threshold values used in the fire detection systems might be modified on a case by case basis. In this paper, an automatic fire detection system without the heuristic fixed threshold values was studied. We presented an automatic method using the statistical color model and the binary background mask. We did the experiment using 600 frames from 6 typical different fire video clips. As the experimental results the proposed method showed a good performance of about average 85% detection rate without false positive, compared with the other methods with the heuristic fixed threshold values. Keywords: Emperical value,Fire detection,RGB,threshold,sensors
Nearest Adjacent Node Discovery Scheme for Routing Protocol in Wireless Senso...IOSR Journals
The broad significance of Wireless Sensor Networks is in most emergency and disaster rescue
domain. The routing process is the main challenges in the wireless sensor network due to lack of physical links.
The objective of routing is to find optimum path which is used to transferring packets from source node to
destination node. Routing should generate feasible routes between nodes and send traffic along the selected path
and also achieve high performance. This paper presents a nearest adjacent node scheme based on shortest path
routing algorithm. It is plays an important role in energy conservation. It finds the best location of nearest
adjacent nodes by involving the least number of nodes in transmission of data and set large number of nodes to
sleep in idle mode. Based on simulation result we shows the significant improvement in energy saving and
enhance the life of the network
Phytochemical, cytotoxic, in-vitro antioxidant and anti-microbial investigati...IOSR Journals
Zizyphus rugosa Lam. (Family: Rhamnaceae), locally known as “Bon Boroi” or as “Jongli Boroi” in Bangladesh generally found as a herb on the hills in bunches on thorny branches of the Zizyphus rugosa trees. Its bark and wood are used medicinally for dysentery in China, India, Laos, Burma, Sri Lanka, Thailand and Vietnam. Phytochemical screening of the Leaf extract of Zizyphus rugosa Lam showed different phytoconstituents including carbohydrates (monosaccharides, reducing and mixed-reducing sugars), alkaloid, glycosides, steroids, tannins and saponin. No flavonoid was detected. In DPPH and NO radical scavenging methods, IC50 was moderately satisfactory. IC50 was found 179.713μg/ml and 769.909μg/ml respectively compare with the reference ascorbic acid (15.707μg/ml and 82.642μg/ml respectively). In LPO (Lipid peroxidation) assay the Leaf fraction extract showed moderate inhibition potentiality (IC50 402.835μg/ml) in comparison to standard drug BHT (IC50 32.94μg/ml). In CUPRAC assays, the fraction was found to possess low Total antioxidant content, good flavonoid, and moderate amounts of phenolics, tannin and alkaloid content. The Leaf fraction extract was found to show good toxicity to Brine Shrimp nauplii, (LC50 212.402μg/ml & LC90 10715.91μg/ml) compare with the reference anticancer drug vincristine sulphate (LC50 2.47μg/ml & LC90 42μg/ml). In the antimicrobial study the fraction showed moderate activity against only one bacterium (Shiggla sonni) while the standard drug Chloramphenicol showed very good zone of inhibition against all five types (Salmonella typhi, Staphylococcus aureus, Shiggla sonni, Salmonella paratyphi, Salmonella grb) of bacteria. These findings provide scientific basis for the use of Zizyphus rugosa Lam. leaf ethanolic extract in traditional medicine in the treatment of aforementioned diseases. The plant also possesses moderate antimicrobial activity, good cytotoxic and good to moderate antioxidant activity.
Spectrophotometric Determination of Drugs and Pharmaceuticals by Cerium (IV) ...IOSR Journals
Simple, sensitive, accurate, and precise spectrophotometric methods for quantitative determination of drugs, viz., Darifenacin (DAR), Esmolol Hydrochloride (ESM), Montelukast Sodium (MON), Sildenafil citrate (SIL),Terbinafine (TER) and Tramadol Hydrochloride (TRA) were developed. The method of each drug depends upon oxidation of drugs by Ce (IV) (Excess) and estimating the amount of unreacted Ce (IV) by amaranth dye at 523nm. The calibration curves obeyed Beer’s law over the concentration range of 1.4-7.0 μg ml-1 (DAR), 2-14 μg ml-1 (ESM), 2-10 μg ml-1 (MON), 20-70 μg ml-1 (SIL), 3-21 μg ml-1 (TER) & 2-14 μg ml-1 (TRA). The methods have been validated in terms of guidelines of ICH and applied to analysis of pharmaceuticals.
As Wireless Sensor Networks are penetrating into the industrial domain, many research opportunities are emerging. One such essential and challenging application is that of node localization. A feed-forward neural network based methodology is adopted in this paper. The Received Signal Strength Indicator (RSSI) values of the anchor node beacons are used. The number of anchor nodes and their configurations has an impact on the accuracy of the localization system, which is also addressed in this paper. Five different training algorithms are evaluated to find the training algorithm that gives the best result. The multi-layer Perceptron (MLP) neural network model was trained using Matlab. In order to evaluate the performance of the proposed method in real time, the model obtained was then implemented on the Arduino microcontroller. With four anchor nodes, an average 2D localization error of 0.2953 m has been achieved with a 12-12-2 neural network structure. The proposed method can also be implemented on any other embedded microcontroller system.
Time Series Forecasting Using Novel Feature Extraction Algorithm and Multilay...Editor IJCATR
Time series forecasting is important because it can often provide the foundation for decision making in a large variety of fields. A tree-ensemble method, referred to as time series forest (TSF), is proposed for time series classification. The approach is based on the concept of data series envelopes and essential attributes generated by a multilayer neural network... These claims are further investigated by applying statistical tests. With the results presented in this article and results from related investigations that are considered as well, we want to support practitioners or scholars in answering the following question: Which measure should be looked at first if accuracy is the most important criterion, if an application is time-critical, or if a compromise is needed? In this paper demonstrated feature extraction by novel method can improvement in time series data forecasting process
Prediction of Extreme Wind Speed Using Artificial Neural Network ApproachScientific Review SR
Prediction of an accurate wind speed of wind farms is necessary because of the intermittent nature
of wind for any region. Number of methods such as persistence, physical, statistical, spatial correlation, artificial
intelligence network and hybrid are generally available for prediction of wind speed. In this paper, ANN based
methods viz., Multi Layer Perceptron (MLP) and Radial Basis Function (RBF) neural networks are used. The
performance of the networks applied for prediction of wind speed is evaluated by model performance indicators
viz., Correlation Coefficient (CC), Model Efficiency (MEF) and Mean Absolute Percentage Error (MAPE).
Meteorological parameters such as maximum and minimum temperature, air pressure, solar radiation and
altitude are considered as input units for MLP and RBF networks to predict the extreme wind speed at Delhi.
The study shows the values of CC, MEF and MAPE between the observed and predicted wind speed (using
MLP) are computed as 0.992, 95.4% and 4.3% respectively while training the network data. For RBF network,
the values of CC, MEF and MAPE are computed as 0.992, 95.9% and 3.0% respectively. The model
performance analysis indicates the RBF is better suited network among two different networks studied for
prediction of extreme wind speed at Delhi.
This paper introduces the Artifi cial Neural Networks (ANN) function to model probabilistic dependencies, in supervised classification tasks for discrimination between earthquakes and explosions problems. ANNs are regarded as the discriminating tools to classify the natural seismic events (earthquakes) from the artifi cial ones (Man-made explosions) based on the seismic signals recorded at regional distances. The bulk of our novel is to improve the obtained numerical results using this advance technique. The ANNs, by testing the different types of seismic features, showed the potential application of this method to discriminate the classes. During the above study, we found out that the Neural Networks have been used in a fully innovative manner in this work. Here the ARMA coefficients filters detects
the type of the source whenever a natural or artificial source changes the nature of the background noise of the seismograms. During the above study, we found out that this algorithm is sometimes capable to alarm the further natural seismological events just a little before the onset.
P REDICTION F OR S HORT -T ERM T RAFFIC F LOW B ASED O N O PTIMIZED W...ijcsit
Short term traffic forecasting has been a very impo
rtant consideration in many areas of transportation
research for more than 3 decades. Short-term traffi
c forecasting based on data driven methods is one o
f the
most dynamic and developing research arenas with en
ormous published literature. In order to improve
forecasting model accuracy of wavelet neural networ
k, an adaptive particle swarm optimization algorith
m
based on cloud theory was proposed, not only to hel
p improve search performance, but also speed up
individual optimizing ability. And the inertia weig
ht adaptively changes depending on X-conditional cl
oud
generator which has the stable tendency and randomn
ess property .Then the adaptive particle swarm
optimization algorithm based on cloud theory was us
ed to optimize the weights and thresholds of wavele
t
BP neural network, Instead of traditional gradient
descent method . At last, wavelet BP neural network
was
trained to search for the optimal solution. Based o
n above theory, an improved wavelet neural network
model based on modified particle swarm optimization
algorithm was proposed and the availability of the
modified prediction method was proved by predicting
the time series of real traffic flow. At last, the
computer simulations have shown that the nonlinear
fitting and accuracy of the modified prediction
methods are better than other prediction methods.
Crude Oil Price Prediction Based on Soft Computing Model: Case Study of IraqKiogyf
Crude Oil Price Prediction Based on Soft Computing Model: Case Study of Iraq
Abstract: The prediction of the price of crude oil is important for economic, political, and industrial purposes in both crude oil importing and exporting countries. Fluctuations in oil prices can have a significant influence in many countries. Therefore, it is necessary to develop a suitable model that can accurately predict different economic and engineering parameters that are directly related to the crude oil price. This paper proposes the use of a soft computing (SC) model which consists of a multi-layer perceptron neural network (MLP-NN) for accurate future crude oil price prediction. The performance of the SC model proposed in this study was compared to that of other neural network approaches and found to perform better in the prediction of both monthly and daily crude oil prices, especially where there is a limited number of input data for model training and in situations of high parameters variability.
Keywords: Crude Oil Price, Soft Computing, Economics, Artificial Neural Networks.
QUAD TREE BASED STATIC MULTI HOP LEACH ENERGY EFFICIENT ROUTING PROTOCOL: A N...IJCNCJournal
This research work propounds a simple graph theory semblance Divide and Conquer Quad tree based Multi-hop Static Leach (DCQMS-Leach) energy efficient routing protocol for wireless sensor networks. The pivotal theme of this research work is to demonstrate how divide and conquer plays a pivotal role in a multi-hop static leach energy efficient routing protocol. This research work motivates, enforces, reckons the DCQMS-Leach energy efficient routing protocol in wireless sensor networks using Mat lab simulator.This research work also computes the performance concepts of DCQMS-Leach routing protocol using various performance metrics such as Packet Drop Rate (PDR), Throughput, and End to End Delay (EED) by comparing and contrasting alive nodes with number of nodes, number of each packets sent to the cluster heads with rounds, number of cluster heads with rounds, number of packets forwarded to the base station with rounds and finally dead nodes with number of rounds. In order to curtail energy consumption this research work proffers a routing methodology such as DCQMS-Leach in energy efficient wireless,sensor routing protocol. The recommended DCQMS-Leach overcomes the in adequacies of all other different leach protocols suggested by the previous researchers.
Elevating Tactical DDD Patterns Through Object CalisthenicsDorra BARTAGUIZ
After immersing yourself in the blue book and its red counterpart, attending DDD-focused conferences, and applying tactical patterns, you're left with a crucial question: How do I ensure my design is effective? Tactical patterns within Domain-Driven Design (DDD) serve as guiding principles for creating clear and manageable domain models. However, achieving success with these patterns requires additional guidance. Interestingly, we've observed that a set of constraints initially designed for training purposes remarkably aligns with effective pattern implementation, offering a more ‘mechanical’ approach. Let's explore together how Object Calisthenics can elevate the design of your tactical DDD patterns, offering concrete help for those venturing into DDD for the first time!
State of ICS and IoT Cyber Threat Landscape Report 2024 previewPrayukth K V
The IoT and OT threat landscape report has been prepared by the Threat Research Team at Sectrio using data from Sectrio, cyber threat intelligence farming facilities spread across over 85 cities around the world. In addition, Sectrio also runs AI-based advanced threat and payload engagement facilities that serve as sinks to attract and engage sophisticated threat actors, and newer malware including new variants and latent threats that are at an earlier stage of development.
The latest edition of the OT/ICS and IoT security Threat Landscape Report 2024 also covers:
State of global ICS asset and network exposure
Sectoral targets and attacks as well as the cost of ransom
Global APT activity, AI usage, actor and tactic profiles, and implications
Rise in volumes of AI-powered cyberattacks
Major cyber events in 2024
Malware and malicious payload trends
Cyberattack types and targets
Vulnerability exploit attempts on CVEs
Attacks on counties – USA
Expansion of bot farms – how, where, and why
In-depth analysis of the cyber threat landscape across North America, South America, Europe, APAC, and the Middle East
Why are attacks on smart factories rising?
Cyber risk predictions
Axis of attacks – Europe
Systemic attacks in the Middle East
Download the full report from here:
https://sectrio.com/resources/ot-threat-landscape-reports/sectrio-releases-ot-ics-and-iot-security-threat-landscape-report-2024/
Generative AI Deep Dive: Advancing from Proof of Concept to ProductionAggregage
Join Maher Hanafi, VP of Engineering at Betterworks, in this new session where he'll share a practical framework to transform Gen AI prototypes into impactful products! He'll delve into the complexities of data collection and management, model selection and optimization, and ensuring security, scalability, and responsible use.
Dr. Sean Tan, Head of Data Science, Changi Airport Group
Discover how Changi Airport Group (CAG) leverages graph technologies and generative AI to revolutionize their search capabilities. This session delves into the unique search needs of CAG’s diverse passengers and customers, showcasing how graph data structures enhance the accuracy and relevance of AI-generated search results, mitigating the risk of “hallucinations” and improving the overall customer journey.
GraphRAG is All You need? LLM & Knowledge GraphGuy Korland
Guy Korland, CEO and Co-founder of FalkorDB, will review two articles on the integration of language models with knowledge graphs.
1. Unifying Large Language Models and Knowledge Graphs: A Roadmap.
https://arxiv.org/abs/2306.08302
2. Microsoft Research's GraphRAG paper and a review paper on various uses of knowledge graphs:
https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/
Climate Impact of Software Testing at Nordic Testing DaysKari Kakkonen
My slides at Nordic Testing Days 6.6.2024
Climate impact / sustainability of software testing discussed on the talk. ICT and testing must carry their part of global responsibility to help with the climat warming. We can minimize the carbon footprint but we can also have a carbon handprint, a positive impact on the climate. Quality characteristics can be added with sustainability, and then measured continuously. Test environments can be used less, and in smaller scale and on demand. Test techniques can be used in optimizing or minimizing number of tests. Test automation can be used to speed up testing.
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...DanBrown980551
Do you want to learn how to model and simulate an electrical network from scratch in under an hour?
Then welcome to this PowSyBl workshop, hosted by Rte, the French Transmission System Operator (TSO)!
During the webinar, you will discover the PowSyBl ecosystem as well as handle and study an electrical network through an interactive Python notebook.
PowSyBl is an open source project hosted by LF Energy, which offers a comprehensive set of features for electrical grid modelling and simulation. Among other advanced features, PowSyBl provides:
- A fully editable and extendable library for grid component modelling;
- Visualization tools to display your network;
- Grid simulation tools, such as power flows, security analyses (with or without remedial actions) and sensitivity analyses;
The framework is mostly written in Java, with a Python binding so that Python developers can access PowSyBl functionalities as well.
What you will learn during the webinar:
- For beginners: discover PowSyBl's functionalities through a quick general presentation and the notebook, without needing any expert coding skills;
- For advanced developers: master the skills to efficiently apply PowSyBl functionalities to your real-world scenarios.
UiPath Test Automation using UiPath Test Suite series, part 5DianaGray10
Welcome to UiPath Test Automation using UiPath Test Suite series part 5. In this session, we will cover CI/CD with devops.
Topics covered:
CI/CD with in UiPath
End-to-end overview of CI/CD pipeline with Azure devops
Speaker:
Lyndsey Byblow, Test Suite Sales Engineer @ UiPath, Inc.
Securing your Kubernetes cluster_ a step-by-step guide to success !KatiaHIMEUR1
Today, after several years of existence, an extremely active community and an ultra-dynamic ecosystem, Kubernetes has established itself as the de facto standard in container orchestration. Thanks to a wide range of managed services, it has never been so easy to set up a ready-to-use Kubernetes cluster.
However, this ease of use means that the subject of security in Kubernetes is often left for later, or even neglected. This exposes companies to significant risks.
In this talk, I'll show you step-by-step how to secure your Kubernetes cluster for greater peace of mind and reliability.
Why You Should Replace Windows 11 with Nitrux Linux 3.5.0 for enhanced perfor...SOFTTECHHUB
The choice of an operating system plays a pivotal role in shaping our computing experience. For decades, Microsoft's Windows has dominated the market, offering a familiar and widely adopted platform for personal and professional use. However, as technological advancements continue to push the boundaries of innovation, alternative operating systems have emerged, challenging the status quo and offering users a fresh perspective on computing.
One such alternative that has garnered significant attention and acclaim is Nitrux Linux 3.5.0, a sleek, powerful, and user-friendly Linux distribution that promises to redefine the way we interact with our devices. With its focus on performance, security, and customization, Nitrux Linux presents a compelling case for those seeking to break free from the constraints of proprietary software and embrace the freedom and flexibility of open-source computing.
The Art of the Pitch: WordPress Relationships and SalesLaura Byrne
Clients don’t know what they don’t know. What web solutions are right for them? How does WordPress come into the picture? How do you make sure you understand scope and timeline? What do you do if sometime changes?
All these questions and more will be explored as we talk about matching clients’ needs with what your agency offers without pulling teeth or pulling your hair out. Practical tips, and strategies for successful relationship building that leads to closing the deal.
Sudheer Mechineni, Head of Application Frameworks, Standard Chartered Bank
Discover how Standard Chartered Bank harnessed the power of Neo4j to transform complex data access challenges into a dynamic, scalable graph database solution. This keynote will cover their journey from initial adoption to deploying a fully automated, enterprise-grade causal cluster, highlighting key strategies for modelling organisational changes and ensuring robust disaster recovery. Learn how these innovations have not only enhanced Standard Chartered Bank’s data infrastructure but also positioned them as pioneers in the banking sector’s adoption of graph technology.
Transcript: Selling digital books in 2024: Insights from industry leaders - T...BookNet Canada
The publishing industry has been selling digital audiobooks and ebooks for over a decade and has found its groove. What’s changed? What has stayed the same? Where do we go from here? Join a group of leading sales peers from across the industry for a conversation about the lessons learned since the popularization of digital books, best practices, digital book supply chain management, and more.
Link to video recording: https://bnctechforum.ca/sessions/selling-digital-books-in-2024-insights-from-industry-leaders/
Presented by BookNet Canada on May 28, 2024, with support from the Department of Canadian Heritage.
DevOps and Testing slides at DASA ConnectKari Kakkonen
My and Rik Marselis slides at 30.5.2024 DASA Connect conference. We discuss about what is testing, then what is agile testing and finally what is Testing in DevOps. Finally we had lovely workshop with the participants trying to find out different ways to think about quality and testing in different parts of the DevOps infinity loop.
Smart TV Buyer Insights Survey 2024 by 91mobiles.pdf91mobiles
91mobiles recently conducted a Smart TV Buyer Insights Survey in which we asked over 3,000 respondents about the TV they own, aspects they look at on a new TV, and their TV buying preferences.
Smart TV Buyer Insights Survey 2024 by 91mobiles.pdf
G013124354
1. IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE)
e-ISSN: 2278-1684,p-ISSN: 2320-334X, Volume 13, Issue 1 Ver. II (Jan. - Feb. 2016), PP 43-54
www.iosrjournals.org
DOI: 10.9790/1684-13124354 www.iosrjournals.org 43 | Page
Determination bearing capacity of driven piles in sandy soils
usingArtificial Neural Networks (ANNs)
1
Ahmadreza Mazaheri,2
Mazaher Berneti
1
Department of Engineering, Mahallat Institute of Higher Education
2
Department of Civil Engineering, TarbiatModarres University
Abstract:Due to some effective factors such as inhomogeneity of soil environment surrounded one pile,
implementation of pile, pile material and its shape, correct estimation of pile load carrying capacity is difficult.
However, pile load test with a high accuracy can be used as a reliable method in different steps of design and
analysis, it is costly and its implementation takes long time in civil projects. This imposed some limitation on
using this test. In contrast, modeling by artificial neural networks is a method which doesn’t require
simplification and use of high reliability coefficients and is based on pervious data and information.
In this paper, artificial neural network is applied for prediction of load carrying capacity of open ended metal
piles pounded in sandy soils. Four parameters of pile length, its diameter, soil elastic modulus and soil internal
friction angle are used as input and pile load carrying capacity is applied as output. How to designa network
and effective factors for its behavior are studied in the related topic. Finally, the effects of four input parameters
on network output are discussed using sensitivity analysis of introduced optimized structure.
Key words:pile load test, load carrying capacity determination device, artificial neural networks, Multi-Layer
Perceptron, sensitivity analysis.
I. Introduction
In the design of pile foundations a good estimation of the pile bearing capacity is one of the major
concerns. Bearing capacity designs are carried out by determining the allowed pile load, which is obtained by
dividing the ultimate pile load by an assumed factor of safety. There are several static and dynamic load test
based approaches to specify the pile capacity, But due to practical problems and the financial considerations,
researchers are forced to look into the alternate numerical methods to determine the pile capacities.One of the
alternate ways to reduce the time and costs are using soft computing techniques includes Artificial Neural
Networks (ANN) and Fuzzy Systems (FS), which can be used to predict the load bearing capacity of piles using
both static and dynamic data sets. (Chan et al., 1995 ; Lee et al., 1996 ; Das et al., 2006 ; Padmini et al., 2008).
Fuzzy logic set theoretic models try to imitate human reasoning and the capability of handling uncertainty
whereas neural network models attempt to emulate architecture and information representation scheme of
human brain (Jang and Sun 1998). Hence neuro fuzzy computing acts as more intelligent systems. Artificial
neural network is used for learning and adaptation whereas fuzzy systems are used to supplement its application
domain. (Nasira et al., 2008). Back-propagation neural network (BPNN) , developed by Ramelhart et al. (1986),
is one of the most frequentlyused algorithms in artificial neural networks. Tak Kim et al. (2001) used BPNN to
predict the lateral behavior of single and group piles. Padmini et al. (2008) used Adaptive NeuroFuzzy Inference
System (ANFIS) to undertake a comparative study with ANNs and fuzzy inference system (FIS) in predicting
the ultimate bearing capacity of shallow foundations. The results indicate that the ANFIS model was able to
yield a sufficient prediction of the ultimate bearing capacity of shallow foundations and remarkably outperforms
the ANN and FIS models although all three models comes more efficientthan the theoretical methods.In recent
years there were number of studies using ANNs to predict the lateral and axial bearing capacities of pile
foundations in compression or uplift, including driven piles and drilled shafts (Shahin and Jaksa 2006; Pal 2006;
Ahmad et al.,2007; Ardalanetal., 2009; Shahin, 2010; Alkrooshand and Nikraz, 2011; Tarawneh, 2013 ;
Fatehnia et al 2014 ; Goh et al., 2005; Shahin, 2010).Multilayer perceptrons (MLPs) which have been used in
the present study is an algorithmtodetermine the driven piles bearing capacity. MLP is a feedforward artificial
neural network model consists a number of processing elements or nodes that are arranged in layers: an input
layer, an output layer, and one or more inter-mediate layers called hidden layers.Ornek et al. (2012) used MLP
and multilinear regression (MLR) to predict bearing capacity of a circular footing over clay soil. They research
states ANNs as a simple and reliable tool for the bearing capacity of circular footings clay soil and the results
produced high coefficients of correlation for the training and testing data. camparison of the results obtained
from both mothods indicated that MLP produced more accurate results than the MLR technique.M.A. Shahin
(2014) modeled Load-settlement for axially loaded driven piles using recurrent neural networks (RNNs)
developed for any soil type. RNNs has been proposed by Jordan (1986) imply an extension of the MLPs with
current-state units, which are processing elements that remember past activity. Based on this study RNN model
2. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 44 | Page
can capture the highly non-linearload-settlement response of steel driven piles reasonably well.This study will
present use of artificial neural networks (ANNs) and the Multi-Layer Perceptron model (MLP) to predictbearing
capacities of steel driven piles in sandy soils. The data used to run the network models have been obtained from
several in-situ full-scale pile load testsandgeotechnical investigations in Hormozgandockyard. An MLP
distinguishes itself by the presence of one or more hidden layers, whose computation nodes are correspondingly
called the hidden neurons of hidden units.In this study one hidden layer MLPs is applied.
II. Artificial Neural Networks
Artificial neural network is a set of simple interconnected computational elements called neuron whose
unique learning capability enables this system to learn very complicated relations. These computational units are
connected to many interconnections in which all the knowledge from environment is stored. First neuron model
which is known as basis of present neural networks was introduced by Piths and McCulloch (1943) (Figure 1).
Three main elements are observed in this model:
- Set of synapses or connections which are determined by weight values or resistance. One xjsignal (j th input)
which is connected to kth neuron is multiplied by wkj weigh. Suffix k and j represent the number of considered
neuron and input connection, respectively.
- One adder sums inputs of a neuron (wkjxj) with a bias value. This action is called linear combination.
- One activation function is applied for limiting neuron output to an appropriate range. Activation function or
simulator can be linear or nonlinear. The most common simulator functions arelinear function, Log-Sigmoid
Transfer Function, two-valued symmetric threshold function and Hyperbolic Tangent Sigmoid Transfer
Function.
2-1 Multi-Layer Perceptron neural network
One of the most common neural networks is Multi-Layer Perceptron neural network or briefly MLP
network. This network has an input layer, one output layer and several hidden layers. In this network which for
the first time was presented by Widrow and Rosenblatt neurons in each layer are fully connected toall neurons
in previous and next layers and there is no reversible connection in the network [1]. MPL neural network having
a hidden layer is able to describe every nonlinear relation between input and output patterns [1]. Therefore, a
neural network withone hidden layer is used in this study.
2-1-2 Determining the number of intermediate neurons
Intermediate neurons act as pattern recognizers [2] and let network learn nonlinear mappings well. It
has been shown that MLP networks having a hidden layer will be able to approximate all functions with
alldegrees of approximation by takingderivative of transfer functions in intermediate and output layers provided
that there are enough neurons in hidden layer [3]. In another words, if there is a mapping, one MLP network
having one hidden layer can be determined to approximate this mapping.
Determining the number of neurons in hidden layer has a significant effect on network behavior. If the
number of neurons is few, neural network can’t reflex nonlinear mapping between input and output accurately
(under fitting). On the other hand, if the number of neurons is more than required number, network learns
training data well (memorizing) by producing complex nonlinear mapping but new data don’t have an
appropriate performance and indeed network lose its generalization capability (over fitting). The number of
neurons in hidden layer is usually obtained experimentally [4].
2-1-3 Training and testing network
The purpose of training in MLP network is adjusting network free parameters (weights) in order to
receive appropriate responses from it. Therefore, in training process, inputs from some parts of information bank
(training set) are presented to network. Output values are calculated and are compared with target values and
then weights are corrected based on the errors. Each presentation of inputs and weights adjustment is called
training cycle. Training cycles continues until the error becomes an acceptable value. After training, weight
values are stored and network is testedby other parts of data that has not been used in training process. In this
study, method of standard error back propagation with momentum is used. In addition to increasing training
speed this method also prevents network instability.
III. Field tests
Training artificial neural networks needs using results of tests taken on real pile samples. For this aim
and regarding accurate pile load tests and their acceptable results, results of 21 pile compressive load tests which
were taken for complementary geotechnical studies of dry poolbyPazhooheshOmranRahvar consulting
3. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 45 | Page
engineers in ship building complex of Hormozgan region, are used. Iran Ship Building and Offshore Industries
Complex is located 40 km west of Bandarabas. According to division of sedimentary-building units of Iran, the
considered area is situated in Zagros folded zone.
According to geotechnical studies that performed in 1995, general situation of soil layers in the location
of project is as following:
- Green mixture of silty sand and silty clay with a thickness of ~ 3 m (approximately to depth of 3m)
- Alternation of green silty sand layer with thickness of ~4 m (depth of ~3 to 7 m)
- Stone layer with average thickness of 20 cm (depth of ~7 m)
- Green silty sand layer (some gritty) with thickness of ~2 m (depth of ~7 to 9 m)
- Silty clay layer with mudhaving a thickness of ~8 m (depth of 9 to 17 m)
- Red silty sand to the end of boreholes (()گمانهdepth more than 17 m)
Figure 2 shows one cross-section of considered area.
3-1 Replacing clay by sand
As it is seen in Figure 2, it is obvious that soil of the region has fine-grained soil layers. Fine-grained
layers are as two separate clayey-sandy silt and silty clay whose average SPT values are from 18 to 22 and the
situation of soil from density aspect is estimated to be stiff to very stiff.
Mentioned clay layer influences tested piles in the region and some part of pile load carrying capacity
is influenced by this layer. On the other hand, laboratory pile were tested in sandy beds. Therefore, in order to
make condition of laboratory tests and field tests similar, it is required to replace clay layers of the region by
sandy layers which have the same density as clay layer and produce a capacity equivalent to clay layer capacity.
For this purpose, firstly it is essential to calculate the amount of load which clay layer can bear.
Since tips of all piles pounded in the region, is placed in underneath coarse-grained soil, clay layer produces a
frictional capacity in all piles.
Considering clay layer thickness and other related characteristics, frictional resistance value due to clay
layer was calculated for all piles using three methods of λ،β and α and their average value was considered as
frictional resistance of clay layer.
After determining the amount of clay layer contribution to overall load carrying capacity of pile, the
purpose of next step is finding sand that firstly has the same density as clay layer and secondly shows a
resistance equivalent to clay layer resistance. In this part different sand with various internal friction angles were
considered and finally sandy soil with density of 1.75 T/m3
and internal friction angle of 34° selected for
alternative sand.
It must be mentioned that some errors occurs in obtained resultsbecause of replacing soil for which
experimental equations are used. Regarding the fact that most of resulting capacity in in-situ experiments is
related to load carrying capacity of pile tip which means thatobtained frictional resistance of clay layerforms a
small percent of overall capacity, obtained error is small and ignorable.
IV. Laboratory experiments
4-1 Pile load test
Since the number of taken real tests was limited and on the other hand training the network required
many data, it was required to take several tests in laboratory. In this study for determining load carrying capacity
of vertical piles under axial load, pile load carrying capacity determination testsweretaken using a large scale
load carrying device which was made in soil and foundation laboratory of TarbiatModares University.
Mentioned device consists of three main sections. These three sections includes: sample making tank
(100 cm length ×100 cm wide × 85 cm height), force applicationsection and force and settlementmeasurement
section. Figure 3 shows total view of load carrying capacity device.
Tests on thin wall tubular piles with lengths of 35, 40, 45 and 50 cm, diameters of 20, 25 and 32 mm and
thickness of 0.9 mm were taken in sandy beds having relative densities of 50, 60, 70, 75 and 80 percent. The
total number of tests was 60 tests.
Standard sand called Firouzkouhbroken silica sand which is briefly named standard sand 161 was used.
This type of sand has a golden yellowish color and based on unified soil classification it is classified in SP
category (poorly grained sand). Graining curve for considered sand is shown in Figure 4.
Other characteristics of this sand include:
Grains density 2.66=GS
Soil void ratio in the stiffest condition emax = 0.928
Soil void ration in the densest condition emin= 0.583
Diameter of particles than which half of particles are finer D50 =0.26
Percentagepassing sieve No. 200 F (%)= 0
Uniformity coefficient 𝐶𝑢 = 𝐷60 𝐷10 = 1.8
4. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 46 | Page
Coefficient of Gradation or Curvature 𝐶𝑐 = 𝐷30
2
𝐷60 (𝐷10) = 1.19
Sandy beds samples were made in load carrying capacity tank. In order to make sandy samples having
desired relative densities, specific weight of sand was poured in load carrying capacity tank and each layer was
pounded to the height of 10 cm.
After preparing sandy beds for pounding pile,one guide tube which was connected to adjacent column
of device by a metal arm was used in order to minimize load eccentricity so as piles are placed inside guide tube
and they are pounded.
For applying force, one hydraulic jack having capacity of 5 tons was used. Mentioned hydraulic jack
ends intest plate (which is situated on pile) from bottom and a movable column (made by a 100 mm box) which
is connected to a crane from top. Loading in these tests was done by standard loading method according to
ASTM standard.
Force and settlement measurement section consists of a force meter ring and 3 electrical strain gauges.
For measuring applied force by hydraulic jack, one 6 tons capacity force gauge ring having a strain gauge
indicator timer with accuracy of 0.002 mm was used. The amount of pile’s strain was calculated by three
electrical strain gauges with accuracy of 0.01 mm. these strains are shown as vertexes of one equilateral triangle
on the monitor.
Finally after finishing tests and plotting force-displacement diagrams for each pile, load carrying
capacities were calculated by Chin’s method for all piles. Based on Chin’s method, firstly displacement values
are divided by their equivalent forces and then diagram of displacement versus this value (division quotient of
displacement by force) is plotted. If the obtained results by this method lie on a straight line, the slope of this
line represents load carrying capacity of pile tip. But in most cases obtained results lie on two straight lines,
therefore the slope of first line determines wall load carrying capacity and the second slop represents total pile
load carrying capacity (Figure 5).
4-2 Plate load Test- PLT
Since soil elastic modulus is considered as one input of neural network, some load tests were taken to
obtain these parameters. One circular solid plate with diameter of 20 cm was used. System for application of
compressive loading is similar to that of pile loading tests and three stain gauges were used for determining the
amount of plate settlement due to applied load. In Figure 6, load carrying capacity device is shown while plate is
tested. Force-displacement diagrams were also plotted in these tests and elastic moduli for different densities
were calculated using equation 1.
E =
Q
δD
1 − µ2
∗ I
δ= produced settlement (strain) due to load Q
D = solid plate diameter
µ = Poisson’s coefficient
Q = applied load
I = correction factor related to test condition and arrangement
V. Applied Neural Network in this study
5-1 Information bank
In this study, results of 21 pile compressive load tests were used.
As mentioned previously, 60 compressive load tests were taken in TarbiatModares University using a
large scale load carrying capacity device.
Totally there were 81 data for training, evaluating and testing the network. 51 data were used for
training, 15 data were used for evaluating and 15 data were used for testing.
5-2 input and output parameters
Based on effective factorsforpile load carrying capacity, parameters which are considered as input of
models include:
- Soil properties: in this study, internal friction angle, ø, was used for modeling and determining soil resistive
properties. Also due to importance of soil elastic modulus in load carrying capacity, this parameter is considered
as input parameter.
- Properties of tested pile: since pile dimensions influence on its load carrying capacity, two parameters of pile
diameter (D) and pile length (L) are also considered as inputs of the network.
It must be mentioned that because laboratory tests were taken in very smaller dimensions than real dimensions,
centrifuge scale rule was used for compensating the effect of scale. Based on this method, different parameters
are change according to Table 1 [8]. N is scale number and since selection of 45 for scale number caused
5. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 47 | Page
laboratory tests data to approach in-situ tests data, 45 was selected for N. For example, based on Table 1 and
selection of 45 as scale number, laboratory piles lengths are multiplied by 45 (N) in order to have real
dimensions.
For example, applied inputs and outputs of the network before multiplication by scale for apile with
diameter of 32 mm, length of 45 cm in different sand densities are listed in Table 2.These numbers must be
multiplied by scales mentioned in Table 1 before beingused in MLP network. Since input and output variables
varies in a wide range, it is better to transfer them to [0 , 1] interval or [-1, 1] interval. The most important
purpose of this transfer is correcting distribution of input and output variable so that network modeling error
reduces [2]. Linear transfer is the most common transfer in neural networks. Therefore, in this study input and
output variables were normalized respect to their maximum and minimum values by a linear relation (equation
2).
( )
U B L B
N P S P M in P L B
M a xP M in P
5-3 Evaluation indexes
For comparing models and their evaluation, some indexes, which can judge models performance for
predicting load carrying capacity of piles, are required. Therefore, below indexes were considered in this study:
- Coefficient of correlation (R): this index shows degree of connection between two variables. Cofficient of
correlation for two x and y variables is defined as following [6]:
22
yyxx
yyxx
R
In which x and y
are average values of x and y, respectively. great R values indicate strong connection between
variables in two data sets. In contrast, small R values shows poor or no connection between two sets [6].
- Mean Absolute Error (MAE): represents average value of error in considered set. This index is expressed by
below equation:
1
1
N
i
i
M A E E
N
- Mean Square Error (MAE): this index also shows average value of error, the difference between obtained
results by tests and those by models with this difference that it focuses more on greater errors [7].
2
1
1
( )
N
i
i
M S E E
N
- Maximum Absolute Error (MAXAE): this index indicates maximum error occurred in considered set. This
index is defined by following equation:
( )i
M AXAE M ax E
- Standard Deviation of Absolute Error (SDAE): this index determines variation degree of absolute error around
MAE. It is obvious that the smaller is this index, the nearer is error of model in overall set to average value and
the more stable is model.
5-4 Implementing network – training and stopping training
For implementing training and testing networks, neural network toolbar Matlab R2009b was used and
networks were trained using LonbergMarca (ML) method. In order to increase generalization capability of
network, cross-validation method was used for stopping training. Since the number of intermediate neurons
influence on networks behavior, one study was performed about their performance with different numbers of
intermediate neurons as following:
Evaluation set as one part of not experimented data which are used for controlling training process, can
simultaneously show network simulation capability (against experimented training data) and network prediction
capability (against not experimented test and evaluation data) in initial analyses. Therefore, firstly performance
of trained networks with different intermediate neurons against these data is evaluated based on error indexes.
(In this stage, coefficient of correlation is calculated for training and test sets). For more accurate selection,
networks that have a good performance against evaluation data are also studied against training and test data.
Finally a network having best performance in simulation (against training data) and in prediction (against test
and evaluation sets) is selected as a network having optimized number of neurons.
In this study, different sets with 4, 6, 8, 10, 12, 14, 16, 18, 20, 22 and 24 intermediate neurons were
evaluated. Coefficient of correlation and mean square error index of evaluation data are shown in Table 3 for
MLP network having different neurons. For this aim, each network having constant number of neurons was
tested 20 times and finally a network having minimum error was stored.
6. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 48 | Page
As it is obvious in Table 3, four networks with 6, 8, 10 and 14 neurons in hidden layer that had better
performance than other networks, were selected and were evaluated against training and test data (Table 4).
According to obtained results, the best condition for considered network is use of 6 neurons in hidden layer
because a network with 6 neurons in hidden layer has minimum error values for network evaluation and test
data.
Diagrams of neural network outputsversus target values for training, evaluation and test dataare shown
in Figures 7,8 and 9, respectively. It must be mentioned that considered network in these diagrams has 6
intermediate neurons in hidden layer.
5-5- Effect of different transfer functions on performance
For evaluating performance of different transfer (activation) functions, networks with different transfer
functions in their intermediate and output layers were studied. 6 intermediate neurons were considered for these
networks in hidden layer and other network parameters were the same. In this study, three activation functions
including linear function (pureline), simulator sigmoid function (logsig) and tangent hyperbolic function (tansig)
were considered. It must be mentioned that based on recommendations of experienced people in neural
networks, purelin transfer function was not used in hidden layer [6]. Totally 6 networks with different transfer
functions and other similar parameters can be formed using mentioned activation functions. These networks and
their transfer functions are listed in Table 5. For evaluating performance of these networks, firstly they were
compared using evaluation data.
As it is seen in Table 5, among these 6 networks, two networks whose transfer functions are tansig in
intermediate layer andpureline and tansigin output layer had better performance. Therefore, these two networks
were compared for training and test data using error parameters. Based on error indexes in Table 6, it is
observed that the network whose transfer function are tansig in hidden layer and purelin in output layer has the
best performance.
VI. Sensitivity Analysis
Studying how uncertainties in output model are attributed to different uncertainties sources in input
model is called sensitivity analysis [. Some advantages of sensitivity analysis are informing model creatorof its
technical errors and determining critical areas in inputs space.
6-1 Results analysis and sensitivity analysis
In this study, sensitivity analysis was performed on three inputs including pile length, pile diameter,
sand internal friction angle and sand elastic modulus.
Relative derivatives of output respect to four inputs were separately calculated for each point of data set
which has its relative input. Statistical properties of relative derivative values are listed in Table 7. As it is
expected, relative derivatives of outputs respect to each input in inputs space have different values.
Average value of sensitivities can be considered as resultant of input effect on output. Indeed, changes
in output values due to changes in inputs is influenced by resultant of output changes respect to one variable
unite in inputs or in another word output sensitivity respect to input to each point of input changes route (from
one initial value to a secondary value ). Of course, average comparison similar to sensitivity values must be
considered relatively to be comparable with other average input values.
Based on mentioned facts and variability of sensitivity values, one method is required for sensitivity
analysis that simultaneously shows scattering and possibility of different output sensitivity values respect to one
input in input space. For this purpose, one statistical method of relative sensitivity values which was used by
Luo et al. (2001) was applied [10]. In this method, five statistical percentages (D10, D25, D50, D75, D90) of
outputs relative sensitivity values respect to considered input are calculated. Based on random samples effect of
an increase or decrease in each input on outputs and dominant general trend in overall input spacearedetermined
by this method. Explanation and interpretation of obtained results will be discussed in the following paragraphs.
D10: represents a value for relative sensitivity than which 90% of values are greater and 10% of values are
smaller. Therefore, if this value becomes positive, it shows that the possibility for positive relative sensitivity
values is greater than 90% or in another word, the possibility for increasing output with increasing considered
input is greater than 90%.
- D25, D75 and D90 have the same explanations as D10.
One input with scattered relative sensitivity percentages values around a base line, has minimum effect
on output comparing with one input whose relative sensitivity percentages are scattered far from a base line.
7. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 49 | Page
Therefore, based on statistical percentages values and their distance from base line, influence degree of each
variable on output can be studied and compared with other variables [10].
Statistical percentages values pertaining to relative sensitivities of load carrying capacity values were
calculated for four input parameters of MLP network having 8 neurons in hidden layer and they are shown in
Table 8.
As it is seen in Figure 10, all parameters have positive relative sensitivity value which indicates
increasing effect of these parameters on load carrying capacity, however based on sensitivity values and their
distances from base line, effects of these parameters are different. For example, because of greater values of
relative sensitivity than internal friction angle and their larger distance from base line it can be concluded that
maximum increasing effect on load carrying capacity occurs due to increasing sand internal friction angle.
Generally, based on distances of statistical categories from base line, percentages values and relative
average values, it can be stated that soil internal friction angle, soil elastic modulus, pile diameter and pile length
respectively have maximum increasing effect on piles load carrying capacity
VII. Conclusion
In this study, MLP neural network was used as an alternative method for predicting load carrying
capacity of metal piles pounded in sandy soils. Obtained results revealed that neural network with minimum
error, high speed and learning capacity, has very high efficiency inpredicting load carrying capacity of metal
piles. Specially MLP network having one hidden layer, 6 intermediate neurons,tansig transfer function in hidden
layer and purelintransfer function in output layerhas the best performance. Also based on performed sensitivity
analysis on the best obtained network, all parameters have an increasing effect onload carrying capacity and
according to sensitivity values and their distances from base line, it is concluded that soil internal friction angle,
soil elastic modulus, pile diameter and pile length respectively have maximum effect on load carrying capacity
of piles.
Figure 1- Neuron mathematics model.
Figure 2- plotted geological cross-section
8. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 50 | Page
Figure 3. device of pile bearing capacity
Figure 4. Graining curve of tested sand
Silty Clay
Sandy Clayey Silt
Silty Sand
9. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 51 | Page
Figure 5- One example of pile force –displacement curve presented by Chin
Figure 5- plate load test
Figure 7- obtained outputs of a network having 6 intermediate neurons versus target values for training data.
10. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 52 | Page
Figure 8- obtained outputs of a network having 6 intermediate neurons versus target values for evaluation data.
Figure 9- obtained outputs of a network having 6 intermediate neurons versus target values for test data with 6
intermediate neurons.
Figure 8.Neural network sensitivity analysis.
-1 0 1 2 3 4
D10
D25
D50
D75
D90
11. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 53 | Page
Table 1- Scale factor for different parameters.
Quantity Scale Factor
for static events
Length, L 1/N
Force, F 1/N2
Modul, E 1
Friction Angle, φ 1
Table 2- input and output parameters for a pile with diameter of 32 mm and length of 45 cm before
multiplication by scales.
Input Output
Dr % L (mm) D (mm) E(Kpa) φ Qu (Kg)
50 32 450 53 37 247.9
60 32 450 83 38 360.3
70 32 450 130 39 448.6
75 32 450 160 39.5 552.7
80 32 450 198 40 591.5
Table 3.Comparing errors of networks having different numbers of neurons in hidden layer for evaluation data.
N.ofH.Neurons Error or Validation Subset
R MSE
4 0.9624 0.0036
6 0.9706 0.0023
8 0.9733 0.0018
10 0.9810 0.0015
12 0.9675 0.0029
14 0.9754 0.0027
16 0.9576 0.0047
18 0.9612 0.0034
20 0.9567 0.0054
22 0.9418 0.0075
24 0.9544 0.0061
Table 4. Comparing errors of networks having different numbers of neurons in hidden layer for training and test
sets.
N.of H. Neurons Error or Training Subset
R MSE MAE MAXAE SDAE
6 0.9951 0.0006 0.014 0.099 0.02
8 0.9854 0.0015 0.025 0.127 0.03
10 0.992 0.00073 0.0167 0.1149 0.021
14 0.9962 0.001 0.021 0.1236 0.023
Error or Testing Subset
6 0.9899 0.0008 0.017 0.1277 0.024
8 0.9895 0.0016 0.027 0.1351 0.029
10 0.9824 0.0023 0.029 0.139 0.031
14 0.9851 0.0021 0.023 0.1349 0.025
Table 5.Effect of different transfer functions on network error for evaluation data.
Transfer Functions Testing Subset
Hid.Layer Out.Layer R MSE
Ttansig Tansig 0.9711 0.0022
Tansig Logsig 0.8835 0.032
Tansig Purelin 0.9706 0.0023
Logsig Tansig 0.9514 0.0068
Logsig Logsig 0.9317 0.012
Logsig Purelin 0.9594 0.0045
12. Determination bearing capacity of driven piles in sandy soils using Artificial Neural…
DOI: 10.9790/1684-13124354 www.iosrjournals.org 54 | Page
Table 6.effect of different transfer functions on network error for training ant test data.
Hidden Output Error or Training Subset
Layer Layer R MSE MAE MAXAE SDAE
Tansig Tansig 0.9872 0.0025 0.023 0.1131 0.027
Tansig Purelin 0.9951 0.0006 0.014 0.099 0.02
Error or Testing Subset
Tansig Tansig 0.9863 0.0018 0.019 0.1245 0.024
Tansig Purelin 0.9899 0.0008 0.017 0.1277 0.024
Table 7.Statistical properties of absolute sensitivity with respect to inputs.
Input Max. Min. Mean. Std. Deviation
D 2.1719 1.2495 1.9196 0.2122
L 1.2194 0.1694 0.7792 0.2588
E 2.5505 0.2309 1.7070 0.6054
ф 3.4732 1.8365 3.0514 0.4204
Table 8. Statistical percentages values of relative sensitivity for each input.
D L E ф
D10 1.6117 0.4279 0.7408 2.4068
D25 1.8002 0.6216 1.2619 2.7728
D50 1.9682 0.7721 1.7728 3.1608
D75 2.0862 0.9952 2.2941 3.436
D90 2.1464 1.1185 2.3248 3.4513
Reference
[1] Haykin, s,; “Neural Networks, a Comprehensive foundation”, Prentice Hall, 1999.
[2] Fausett, L. V. (1994). Fundamentals neural networks: Architecture, algorithms, and applications, Prentice-Hall, Inc., Englewood
Cliffs, New Jersey.
[3] Knowels, P., “Predicting Labor Productivity using neural networks”, MSc thesis, University of Alberta, Edmonth, Canada.
[4] Smith, G.N., Probably and statistics in civil engineering, Collins, 1986.
[5] The, C. I., Wong, K. S., Goh, A. T. C., and Jaritngam, s. (1997). “Prediction of pile capacity using neural network.”.
[6] B. H. Fellenius; “ What Capacity Value to Choose from the Results a static Loading Test”: Winter 2001.
[7] Lu, M., Abourizk, S.M. & Hermann, U.H., Sensitivity Analisys of neural networks in spool fabrication productivity studies, j.
comp. in civil engineerings, Vol. 15, No.4
[8] Chan, W. T., Chow, Y. K., and Liu, I. F. 1995. “Neural networks:An alternative to the pile driving formulas.” Comput. Geotech.,
17(2),135–156.
[9] Lee, I.M., Lee, J.H., 1996. “Prediction of pile bearing capacity using artificial neural networks.” Comput. Geotech. 18(3),189–200.
[10] Das, S.K.,Basudhar,P.K.,2006. “Undrainedlateralloadcapacityofpilesin clay usingartificial neural network.”
Comput.Geotech.33(8),454–459.
[11] Padmini, D., Ilamparuthi, K., Sudheer, K.P., 2008 “Ultimate bearing capacity prediction of shallow foundations on cohesionless
soils using neurofuzzy models.” Comput.Geotech. 35 33–46
[12] J.Jang, C. Sun, E. Mizutani, 1998 “Neuro Fuzzy and Soft Computing”, Prentice Hall, 35-68
[13] Nasira, G. M.,Ashok Kumar, S., SweetyKiruba, M.,2008.“A comparative study of fuzzy logic with artificial neural networks
algorithms in clustering”J. of Comput.Applicat. 1(4), 6-8
[14] Shahin, M.A., Jaksa, M.B., 2006. “Pull out capacity of small ground anchors by direct conepenetration test methods and neural
networks.”Can.Geotech. J. 43(6),626-637.
[15] Pal, M., 2006. “Support vector machines‐based modelling of seismic liquefaction potential”Int. J. Numer. Anal. Methods
Geomech.30(10),983–996.
[16] Ahmad, I., ElNaggar, H., Kahn, A.N., 2007. “Artificial neural network application to estimate
kinematicsoilpileinteractionresponseparameters.”Soil Dyn.Earthq.Eng.27(9),892-905
[17] Ardalan, H., Eslami, A., Nariman-Zadeh, N., 2009. “Piles shaft capacity from CPT and CPTU data by polynomial neural networks
and genetic algorithms.” Comput. Geotech. 36(4), 616-625.
[18] Alkroosh, I., Nikraz, H., 2011. “Simulating pile load-settlement behavior from CPT data using intelligent computing.” Cent. Eur. J.
Eng. 1(3),295-305
[19] Tarawneh, B., 2013. “Pipe pile setup: data base and prediction model using artificial neuralnetwork.” Soils Found. 53(4), 607-615.
[20] Fatehnia, M., Tawfiq, K., Hataf, N., Ozguven, E.E., 2014. “New Method for Predicting the Ultimate Bearing Capacity of Driven
Piles by using Flap Number”KSCE J. Civil. Eng.Pp 1-10
[21] Goh, A.T.C.,Kulhawy, F.H., Chua, C.G., 2005. “Bayesian neural network analysis of undrained side resistance of drilled shafts.” J.
Geotech. Geoenviron. Eng. 131(1), 84-93.
[22] Shahin, M.A.,2010.“Intelligentcomputingformodelingaxialcapacityofpile foundations.” Can. Geotech. J.47(2),230-243.
[23] M.A., Shahin.,2014.“Load–settlement modeling of axially loaded steel driven piles using CPT-based recurrent neural
networks.”Soils Found.54(3),515-522
[24] Jordan, M.I., 1986. “Attractor dynamics and parallelism in a connectionist sequential machine.” In:Proceedings of the 8th
Annual
Conference of the Cognitive Science Society. Amherst, MA,531-546.