Multilayer Feedforward Network Research Articles

Clinical application of modified bag-of-features coupled with hybrid neural-based classifier in dengue fever classification using gene expression data.

Dengue fever detection and classification have a vital role due to the recent outbreaks of different kinds of dengue fever. Recently, the advancement in the microarray technology can be employed for such classification process. Several studies have established that the gene selection phase takes a significant role in the classifier performance. Subsequently, the current study focused on detecting two different variations, namely, dengue fever (DF) and dengue hemorrhagic fever (DHF). A modified bag-of-features method has been proposed to select the most promising genes in the classification process. Afterward, a modified cuckoo search optimization algorithm has been engaged to support the artificial neural (ANN-MCS) to classify the unknown subjects into three different classes namely, DF, DHF, and another class containing convalescent and normal cases. The proposed method has been compared with other three well-known classifiers, namely, multilayer perceptron feed-forward network (MLP-FFN), artificial neural network (ANN) trained with cuckoo search (ANN-CS), and ANN trained with PSO (ANN-PSO). Experiments have been carried out with different number of clusters for the initial bag-of-features-based feature selection phase. After obtaining the reduced dataset, the hybrid ANN-MCS model has been employed for the classification process. The results have been compared in terms of the confusion matrix-based performance measuring metrics. The experimental results indicated a highly statistically significant improvement with the proposed classifier over the traditional ANN-CS model.

Accelerated optimizations of an electromagnetic acoustic transducer with artificial neural networks as metamodels

Abstract. Electromagnetic acoustic transducers (EMATs) are noncontact transducers generating ultrasonic waves directly in the conductive sample. Despite the advantages, their transduction efficiencies are relatively low, so it is imperative to build accurate multiphysics models of EMATs and optimize the structural parameters accordingly, using a suitable optimization algorithm. The optimizing process often involves a large number of runs of the computationally expensive numerical models, so metamodels as substitutes for the real numerical models are helpful for the optimizations. In this work the focus is on the artificial neural networks as the metamodels of an omnidirectional EMAT, including the multilayer feedforward networks trained with the basic and improved back propagation algorithms and the radial basis function networks with exact and nonexact interpolations. The developed neural-network programs are tested on an example problem. Then the model of an omnidirectional EMAT generating Lamb waves in a linearized steel plate is introduced, and various approaches to calculate the amplitudes of the displacement component waveforms are discussed. The neural-network metamodels are then built for the EMAT model and compared to the displacement component amplitude (or ratio of amplitudes) surface data on a discrete grid of the design variables as the reference, applying a multifrequency model with FFT (fast Fourier transform)/IFFT (inverse FFT) processing. Finally the two-objective optimization problem is formulated with one objective function minimizing the ratio of the amplitude of the S0-mode Lamb wave to that of the A0 mode, and the other objective function minimizing as the negative amplitude of the A0 mode. Pareto fronts in the criterion space are solved with the neural-network models and the total time consumption is greatly decreased. From the study it could be observed that the radial basis function network with exact interpolation has the best performance considering its accuracy of approximation and the time required to build the metamodel.

Measurement of CO2 absorption enthalpy and heat capacity of aqueous hexamethylenediamine and its aqueous mixture with N-methyldiethanolamine

In this work, enthalpy of carbon dioxide (CO2) absorption in aqueous hexamethylenediamine (HMDA) and its blend with N-methyldiethanolamine (MDEA) is measured using an automated reaction calorimeter in the temperature range of (30–60)°C. Concentrations of aqueous HMDA solution studied are (10, 20, 30) mass% and that for mixed amine system are (5% HMDA+25% MDEA), (10% HMDA+20% MDEA) and (15% HMDA+15% MDEA). The heat capacity of the unloaded aqueous amine solvents is also measured in the same range of temperature and concentration. The experimental enthalpy of absorption data is presented as a function of CO2 loading (mol CO2 per mol total amine) and temperature. A combined model for the prediction of CO2 absorption enthalpy in aqueous HMDA and (HMDA+MDEA) solvent systems is presented using artificial neural network (ANN) tool. A multi-layer feed-forward network is developed using Levenberg-Marquardt back propagation algorithm. Average absolute deviation between experimental and ANN model predicted data is found to be 0.38%.

Application of artificial neural networks as a predictive method to differentiate the wood of Pinus sylvestris L. and Pinus nigra Arn subsp. salzmannii (Dunal) Franco

The wood structure of conifers in general and the Pinus genus in particular makes species differentiation by traditional qualitative or quantitative methods complicated or even impossible at times. Pinus sylvestris L. and Pinus nigra Arn subsp. salzmannii (Dunal) Franco are a clear example of this because they cannot be differentiated by traditional methods. However, correctly identifying these species is very important in some cases as they are extensively used in a large variety of fields because of their wide distribution range in the forests of Europe and Asia. Using trees selected from the same forest to minimise the influence of site and performing a biometric study of 10 growth rings from the same climate period, a feedforward multilayer perceptron network trained by the resilient backpropagation algorithm was designed to determine whether the network could be used to differentiate these species with a high degree of probability. The artificial neural network achieved 90.4% accuracy in the training set, 81.6% in the validation set and 81.2% in the testing set. This result justifies the use of this tool for wood identification at anatomical level.

Prediction of Ore Quantity Based on GA-BP Neural Network

BP neural network is a multilayer feedforward network trained by error back-propagation algorithm, which is one of the most widely used neural network models. However, BP neural network has exposed more and more shortcomings and deficiencies with the expansion of the application scope. In the prediction of ore quantity, BP neural network has the characteristics of slow convergence and easy to fall into local minimum point. In order to obtain the global optimal solution, and to improve the defects of BP neural network, this paper proposes combination optimization algorithm of genetic algorithm (GA) and BP neural network to improve the speed and accuracy of forecasting the main design flow chart and the analysis of the sort distinguish algorithm are offered, and then some problem in the design and debugging of the algorithm are discussed. On this basis, the GA-BP neural network model is constructed and applied to optimize the initial weights and threshold value of BP neural network. This model choices the floating point coding method to encode the connection weights and thresholds, and divides subjects into several populations. Through the introduction of selection, mutation, crossover, initial weight and other operators, making operational synergies between the various groups. This study selects 30 geological units, 8 quantitative variables (Pb, Zn, Cu, Mo, Si, Ni, Co, V) and 12 qualitative variables to carry out empirical analysis. Then the simulation of the algorithm is carried out in MATLAB and the parameters are analysed. By normalizing the input samples, 22 groups of observation data are used as the training data for prediction, and the latter 8 groups of observation data are used as the test data to be verified. The results show that when the ore quantity characteristics are not very significant, the model will produce prediction bias. But the improvement of the algorithm increases the efficiency of the function approach capacity of BP neural network and conquer the BP neural network system’s instability. It provides an auxiliary guide for ore prediction, which have higher reference value.

Detecting redirection spam using multilayer perceptron neural network

Quality information retrieval from Web is essential for every search engine. But the quality of information is being exploited by spammers who make heavy use of malicious redirections for the purpose of phishing, downloading malware or attaining high search engine ranking. Malicious redirections present the irrelevant content to search user, thereby affecting user satisfaction. It also leads to wastage of network bandwidth. In this paper, we propose a neural framework for detecting redirection spam. We incorporated the feed-forward multilayer perceptron network and used scaled conjugate gradient algorithm that is able to perform very fast classification of URLs leading to redirection spam. We investigated the network empirically to choose the number of hidden layers and observed that when network is trained with two hidden layers, it gives better accuracy. We validated our proposed approach against the dataset of 2383 URLs and were able to detect the spammed redirections with high accuracy. The results indicate that neural networks are very effective technique to model the redirection spam detection.

Modelling of input-output relationships of metal inert gas welding process using soft computing-based approaches

Input-output relationships of metal inert gas welding process were determined in both forward and reverse directions, which are required in order to automate the same. Various types of neural networks, namely multi-layer feed-forward network, counter-propagation network and radial basis function network had been used for the said purpose. The networks were trained using back-propagation algorithm and/or genetic algorithm. Their performances were compared, and radial basis function network developed using the concept of clustering was found to perform better than other networks in terms of accuracy in prediction.

Factors influencing the development of visceral metastasis of breast cancer: A retrospective multi-center study

PurposeVisceral metastasis of breast cancer (BC) is an alarming development and correlates with poor median overall survival. The purpose of this retrospective study is to examine the risk factors for developing visceral metastasis by considering tumor biology and patient characteristics. MethodsUsing the BRENDA database, the risk factors such as histological and intrinsic subtypes of BC, age at primary diagnosis, grading, nodal status, tumor size and year of primary diagnosis were examined in univariate and multivariate analysis. Categorical variables were compared by using χ2 tests. Furthermore, multivariate Cox proportional hazards regression models, Kaplan–Meier product-limit method and log-rank test were applied. The results of two tree-building algorithms, “exhausted CHAID” (Chi-squared Automatic Interaction Detector) and CART (Classification and Regression Trees) were verified with further multivariate analysis, radial basis function networks (RBF-net), feedforward multilayer perceptron networks (MLP) and logistic regression. ResultsIn a patient collective of 886 metastasized patients, 56.9% had developed visceral metastases and 27.1% visceral-only metastases. The different histological and intrinsic subtypes of BC and the grading correlate significantly with the visceral-only metastasis behavior, whereas the age at primary diagnosis, the nodal status, the tumor size and the year of the primary diagnosis had no influence. Patients with ductal/other BC, LuminalB/HER2, TNBC, HER2 overexpressing subtype and grade 3 had an increased risk for the development of visceral-only metastasis. ConclusionsIntrinsic and histological subtypes as well as the grading of BC affected significantly the visceral metastasis behavior.

Detection of Urban Damage Using Remote Sensing and Machine Learning Algorithms: Revisiting the 2010 Haiti Earthquake

Remote sensing continues to be an invaluable tool in earthquake damage assessments and emergency response. This study evaluates the effectiveness of multilayer feedforward neural networks, radial basis neural networks, and Random Forests in detecting earthquake damage caused by the 2010 Port-au-Prince, Haiti 7.0 moment magnitude (Mw) event. Additionally, textural and structural features including entropy, dissimilarity, Laplacian of Gaussian, and rectangular fit are investigated as key variables for high spatial resolution imagery classification. Our findings show that each of the algorithms achieved nearly a 90% kernel density match using the United Nations Operational Satellite Applications Programme (UNITAR/UNOSAT) dataset as validation. The multilayer feedforward network was able to achieve an error rate below 40% in detecting damaged buildings. Spatial features of texture and structure were far more important in algorithmic classification than spectral information, highlighting the potential for future implementation of machine learning algorithms which use panchromatic or pansharpened imagery alone.

Investigation of Two Neural Network Methods in an Automatic Mapping Exercise

This paper investigates the performance of two neural network (NN) methods viz. a radial basis function network (RBFN) and a multilayer feed forward network (MFFN) to predict the radioactivity levels at a given test site. A comparative evaluation of the two networks is done using Root mean square error (RMSE), Pearsons r, Mean error (ME) and Mean Absolute error (MAE). It was found that the RBFN performed marginally better compared to the other method

Comparison of linear regression and artificial neural network model of a diesel engine fueled with biodiesel-alcohol mixtures

This study deals with usage of linear regression (LR) and artificial neural network (ANN) modeling to predict engine performance; torque and exhaust emissions; and carbon monoxide, oxides of nitrogen (CO, NOx) of a naturally aspirated diesel engine fueled with standard diesel, peanut biodiesel (PME) and biodiesel-alcohol (EME, MME, PME) mixtures. Experimental work was conducted to obtain data to train and test the models. Backpropagation algorithm was used as a learning algorithm of ANN in the multilayered feedforward networks. Engine speed (rpm) and fuel properties, cetane number (CN), lower heating value (LHV) and density (ρ) were used as input parameters in order to predict performance and emission parameters. It was shown that while linear regression modeling approach was deficient to predict desired parameters, more accurate results were obtained with the usage of ANN.

A comparative study of 3D FZI and electrofacies modeling using seismic attribute analysis and neural network technique: A case study of Cheshmeh-Khosh Oil field in Iran

Abstract Electrofacies are used to determine reservoir rock properties, especially permeability, to simulate fluid flow in porous media. These are determined based on classification of similar logs among different groups of logging data. Data classification is accomplished by different statistical analysis such as principal component analysis, cluster analysis and differential analysis. The aim of this study is to predict 3D FZI (flow zone index) and Electrofacies (EFACT) volumes from a large volume of 3D seismic data. This study is divided into two parts. In the first part of the study, in order to make the EFACT model, nuclear magnetic resonance (NMR) log parameters were employed for developing an Electrofacies diagram based on pore size distribution and porosity variations. Then, a graph-based clustering method, known as multi resolution graph-based clustering (MRGC), was employed to classify and obtain the optimum number of Electrofacies. Seismic attribute analysis was then applied to model each relaxation group in order to build the initial 3D model which was used to reach the final model by applying Probabilistic Neural Network (PNN). In the second part of the study, the FZI 3D model was created by multi attributes technique. Then, this model was improved by three different artificial intelligence systems including PNN, multilayer feed-forward network (MLFN) and radial basis function network (RBFN). Finally, models of FZI and EFACT were compared. Results obtained from this study revealed that the two models are in good agreement and PNN method is successful in modeling FZI and EFACT from 3D seismic data for which no Stoneley data or NMR log data are available. Moreover, they may be used to detect hydrocarbon-bearing zones and locate the exact place for producing wells for the future development plans. In addition, the result provides a geologically realistic spatial FZI and reservoir facies distribution which helps to understand the subsurface reservoirs heterogeneities in the study area.

Evaluation of clinical parameters influencing the development of bone metastasis in breast cancer.

BackgroundThe development of metastases is a negative prognostic parameter for the clinical outcome of breast cancer. Bone constitutes the first site of distant metastases for many affected women. The purpose of this retrospective multicentre study was to evaluate if and how different variables such as primary tumour stage, biological and histological subtype, age at primary diagnosis, tumour size, the number of affected lymph nodes as well as grading influence the development of bone-only metastases.MethodsThis retrospective German multicentre study is based on the BRENDA collective and included 9625 patients with primary breast cancer recruited from 1992 to 2008. In this analysis, we investigated a subgroup of 226 patients with bone-only metastases. Association between bone-only relapse and clinico-pathological risk factors was assessed in multivariate models using the tree-building algorithms “exhausted CHAID (Chi-square Automatic Interaction Detectors)” and CART(Classification and Regression Tree), as well as radial basis function networks (RBF-net), feedforward multilayer perceptron networks (MLP) and logistic regression.ResultsMultivariate analysis demonstrated that breast cancer subtypes have the strongest influence on the development of bone-only metastases (χ2 = 28). 29.9 % of patients with luminal A or luminal B (ABC-patients) and 11.4 % with triple negative BC (TNBC) or HER2-overexpressing tumours had bone-only metastases (p < 0.001). Five different mathematical models confirmed this correlation. The second important risk factor is the age at primary diagnosis. Moreover, BC subcategories influence the overall survival from date of metastatic disease of patients with bone-only metastases. Patients with bone-only metastases and TNBC (p < 0.001; HR = 7.47 (95 % CI: 3.52–15.87) or HER2 overexpressing BC (p = 0.007; HR = 3.04 (95 % CI: 1.36–6.80) have the worst outcome compared to patients with luminal A or luminal B tumours and bone-only metastases.ConclusionThe bottom line of different mathematical models is the prior importance of subcategories of breast cancer and the age at primary diagnosis for the appearance of osseous metastases. The primary tumour stage, histological subtype, tumour size, the number of affected lymph nodes, grading and NPI seem to have only a minor influence on the development of bone-only metastases.

Modelling of Multi Layer Feed forward Neural Networks to Determine the Compressive Strength of Marmara Region Aggregate's Concrete

We aim to estimate concrete compressive strength by using the physical properties of the aggregates that are the main components of concrete. For this aim, concrete samples were prepared with aggregates having different origins and characteristics obtained from different 10 locations of the Marmara region. The compressive strength's results obtained by changing the aggregates were compared by ensuring the other components forming the concrete remained constant. 330 separate experiments were conducted to determine the physical characteristics of the aggregates, and these characteristics were used as input data in a multi layer Feed Forward Network model. The compressive strength of 7 and 28 days of the concrete obtained from the experiments were used as output in to a Multi-layer Feed Forward model. The training and test results from the models coincided closely with experimental results; also, the results were compared to the estimations made with a linear regression method.

Estimation of NMR Total and Free Fluid Porosity from Seismic Attributes Using Intelligent Systems: A Case Study from an Iranian Carbonate Gas Reservoir

Nuclear magnetic resonance (NMR) log is a powerful tool for exploration and development of oil and gas fields since it can be applied to evaluate reservoir and nonreservoir horizons. Total porosity and free fluid porosity are two valuable outputs of NMR log which are accessible through processing of raw logs. In the present study, an attempt has been made to create a quantitative correlation between output parameters of NMR log and seismic attributes using linear regression method and artificial intelligent systems. An integration of 3D seismic data and well log data has been done to predict parameters of NMR log from a new source of data which is spread in the entire field and accessible in the primary stages of field development. For this purpose, the best seismic attributes were selected after extraction of acoustic impedance and sample- based attributes of 3D seismic data using stepwise linear regression method. Multivariate linear regression equations and correlation coefficients with target logs were determined. Finally, three different artificial intelligence systems including probabilistic neural network (PNN), multilayer feed-forward network (MLFN) and radial basis function network (RBFN) were designed and optimized. Results of correlation coefficients between real and predicted logs and also prediction error in the blind test showed that PNN performed better than MLFN and RBFN. At the last step, PNN was used to reconstruct the 3D model of NMR total porosity and free fluid porosity in the reservoir zone of the studied carbonate gas field in the south of Iran.

Development of correlations and artificial neural network models to predict second-degree burn time for thermal-protective fabrics

Firefighters and operators working near industrial furnaces or ovens usually encounter extreme environmental conditions. Heat transfer through fabrics exposed to such high heat flux exposures causes skin burn. There are various parameters which govern heat transfer and second degree burn time in such situations. In this work, six dimensionless parameters associated with heat transfer through fabrics exposed to various flame and radiant heat flux exposures are obtained. Correlations for dimensionless second degree burn time are obtained in terms of other non-dimensional parameters. These separate correlations for each type of exposure can be used to obtain thermal protective performance (TPP) of any fabric exposed to different type of heat conditions, provided that fabric material, fabric geometric parameters, air gap between fabric and skin, and intensity of heat exposure are known. Artificial neural network (ANN) approach is also used to predict TPP of clothing. Multi-layer feed-forward back-propagation networks are developed for different type of exposures. Results shows that both correlations and ANN approach used in the study are promising. ANN model predictions are found to agree better with experimental observations as compared to the outcome of developed correlations. Importance of various dimensionless parameters obtained using dimensional analysis are also emphasized.

Learning Precise Spike Train-to-Spike Train Transformations in Multilayer Feedforward Neuronal Networks.

We derive a synaptic weight update rule for learning temporally precise spike train-to-spike train transformations in multilayer feedforward networks of spiking neurons. The framework, aimed at seamlessly generalizing error backpropagation to the deterministic spiking neuron setting, is based strictly on spike timing and avoids invoking concepts pertaining to spike rates or probabilistic models of spiking. The derivation is founded on two innovations. First, an error functional is proposed that compares the spike train emitted by the output neuron of the network to the desired spike train by way of their putative impact on a virtual postsynaptic neuron. This formulation sidesteps the need for spike alignment and leads to closed-form solutions for all quantities of interest. Second, virtual assignment of weights to spikes rather than synapses enables a perturbation analysis of individual spike times and synaptic weights of the output, as well as all intermediate neurons in the network, which yields the gradients of the error functional with respect to the said entities. Learning proceeds via a gradient descent mechanism that leverages these quantities. Simulation experiments demonstrate the efficacy of the proposed learning framework. The experiments also highlight asymmetries between synapses on excitatory and inhibitory neurons.

Artificial neural networks as classification and diagnostic tools for lymph node-negative breast cancers

Artificial neural networks (ANNs) can be used to develop a technique to classify lymph node negative breast cancer that is prone to distant metastases based on gene expression signatures. The neural network used is a multilayered feed forward network that employs back propagation algorithm. Once trained with DNA microarray-based gene expression profiles of genes that were predictive of distant metastasis recurrence of lymph node negative breast cancer, the ANNs became capable of correctly classifying all samples and recognizing the genes most appropriate to the classification. To test the ability of the trained ANN models in recognizing lymph node negative breast cancer, we analyzed additional idle samples that were not used beforehand for the training procedure and obtained the correctly classified result in the validation set. For more substantial result, bootstrapping of training and testing dataset was performed as external validation. This study illustrates the potential application of ANN for breast tumor diagnosis and the identification of candidate targets in patients for therapy.

Weak signal detection and propagation in diluted feed-forward neural network with recurrent excitation and inhibition

Reliable signal propagation across distributed brain areas provides the basis for neural circuit function. Modeling studies on cortical circuits have shown that multilayered feed-forward networks (FFNs), if strongly and/or densely connected, can enable robust signal propagation. However, cortical networks are typically neither densely connected nor have strong synapses. This paper investigates under which conditions spiking activity can be propagated reliably across diluted FFNs. Extending previous works, we model each layer as a recurrent sub-network constituting both excitatory (E) and inhibitory (I) neurons and consider the effect of interactions between local excitation and inhibition on signal propagation. It is shown that elevation of cellular excitation–inhibition (EI) balance in the local sub-networks (layers) softens the requirement for dense/strong anatomical connections and thereby promotes weak signal propagation in weakly connected networks. By means of iterated maps, we show how elevated local excitability state compensates for the decreased gain of synchrony transfer function that is due to sparse long-range connectivity. Finally, we report that modulations of EI balance and background activity provide a mechanism for selectively gating and routing neural signal. Our results highlight the essential role of intrinsic network states in neural computation.

Particle swarm optimization trained neural network for structural failure prediction of multistoried RC buildings

Faulty structural design may cause multistory reinforced concrete (RC) buildings to collapse suddenly. All attempts are directed to avoid structural failure as it leads to human life danger as well as wasting time and property. Using traditional methods for predicting structural failure of the RC buildings will be time-consuming and complex. Recent research proved the artificial neural network (ANN) potentiality in solving various real-life problems. The traditional learning algorithms suffer from being trapped into local optima with a premature convergence. Thus, it is a challenging task to achieve expected accuracy while using traditional learning algorithms to train ANN. To solve this problem, the present work proposed a particle swarm optimization-based approach to train the NN (NN-PSO). The PSO is employed to find a weight vector with minimum root-mean-square error (RMSE) for the NN. The proposed (NN-PSO) classifier is capable to tackle the problem of predicting structural failure of multistoried reinforced concrete buildings via detecting the failure possibility of the multistoried RC building structure in the future. A database of 150 multistoried buildings’ RC structures was employed in the experimental results. The PSO algorithm was involved to select the optimal weights for the NN classifier. Fifteen features have been extracted from the structural design, while nine features have been opted to perform the classification process. Moreover, the NN-PSO model was compared with NN and MLP-FFN (multilayer perceptron feed-forward network) classifier to find its ingenuity. The experimental results established the superiority of the proposed NN-PSO compared to the NN and MLP-FFN classifiers. The NN-PSO achieved 90 % accuracy with 90 % precision, 94.74 % recall and 92.31 % F-Measure.