Scaled Conjugate Gradient Backpropagation Research Articles

Artificial Neural Network to Detect Human Hand Gestures for a Robotic Arm Control.

Assistive technology is critical to improving daily life of those with muscular issues such as Cerebral Palsy and Duchenne Muscular Dystrophy by augmenting their activities of daily living (ADL). Robotic manipulators are one solution for helping with ADL; however, intuitive, accurate interfaces for higher degrees of freedom (DOF) robotic arms are still lacking. An intuitive control system based on artificial neural network (ANN) classification of real-time surface electromyography (sEMG) signals from the user's forearm to detect nine hand gestures and control the movement of the 6 DOF robotic arm is proposed in this paper. The regular machine learning classifiers with the highest classification accuracies were ensemble-bagged trees at 90.3% and cubic SVM at 89.6%, with linear SVM being 84.8%. However, the classifier chosen was a scaled conjugate gradient backpropagation neural network model, with a classification accuracy of 85%, due to accuracy and usability in a Simulink model. The trained ANN model was incorporated into the control system for the robotic arm and tested in virtual environment. Preliminary testing of the robotic arm shows that the forward kinematic control system works well for most hand poses. Future improvements will include more processing of the sEMG signals and training on sEMG data from multiple subjects for a generalized ANN model.

Hybrid model based on neural networks, type-1 and type-2 fuzzy systems for 2-lead cardiac arrhythmia classification

This paper describes an approach using computational intelligence methods to form a hybrid model as a classification method for 2-lead cardiac arrhythmias. The hybridization of methods can increase the performance in a system and take advantage of the benefits offered by such techniques in solving complex problems. The interpretation of electrocardiograms is a useful task for physicians, but when it comes to reviewing more than 24 h of information, it becomes a laborious task for them. For this reason, the design a computational model that helps in such a task is very useful for the timely medical diagnosis. The hybrid model is build using artificial neural networks and fuzzy logic. Training and testing of the hybrid model was with the Massachusetts Institute of Technology and Beth Israel Hospital (MIT-BIH) arrhythmia database. The heartbeats are preprocessed to improve results of classification. Ten different classes of normal and arrhythmia signals for building the hybrid model are considered. We used two electrode signals or leads included in the MIT-BIH arrhythmia database, MLII and V1, V2, or V3 as second electrode signal. The hybrid model is composed by two basic module units, as described below. A basic module unit to perform the classification for each signal lead is used. Each basic module unit is composed of three different classifiers based on the following models: fuzzy KNN algorithm, multilayer perceptron with gradient descent and momentum (MLP-GDM), and multilayer perceptron with scaled conjugate gradient backpropagation (MLP-SCG). The outputs from the classifiers are combined using a fuzzy system for integration of results. We designed two fuzzy systems, Mamdani type-1 fuzzy system (type-1 FIS) and an interval type-2 fuzzy system (IT2FIS). The reason is to perform a comparison between type-1 FIS and IT2FIS in the hybrid model. We have obtained best results in the classification rate using IT2FIS instead of type-1 FIS in the basic units. Finally, a type-1 FIS is used to determine the global classification for the 2 basic units in hybrid model. We obtained a good classification rate in each basic module unit, 92.90% and 92.70% of classification rate for basic modules unit 1 and unit 2 respectively. Finally, we obtained a 93.80% when used type-1 FIS and 94.20% of classification rate used IT2FIS combining both basic module units. In the results presented, we improve the global classification in proposed hybrid model combining neural networks and fuzzy logic used both signal lead included in MIT-BIH arrhythmia database. The proposed hybrid model maybe extended to use multi-lead arrhythmia classification using other databases that contain 12 leads to be able to make a complete medical diagnosis.

Artificial Neural Network for in-Bed Posture Classification Using Bed-Sheet Pressure Sensors.

Pressure ulcer prevention is a vital procedure for patients undergoing long-term hospitalization. A human body lying posture (HBLP) monitoring system is essential to reschedule posture change for patients. Video surveillance, the conventional method of HBLP monitoring, suffers from various limitations, such as subject's privacy, and field-of-view obstruction. We propose an autonomous method for classifying the four state-of-the-art HBLPs in healthy adults subjects: supine, prone, left and right lateral, with no sensors or cables attached on the body and no constraints imposed on the subject. Experiments have been conducted on 12 healthy adults (age 27.35 ± 5.39 years) using a collection of textile pressure sensors embedded in a cover placed under the bed sheet. Histogram of oriented gradients and local binary patterns were extracted and fed to a supervised artificial neural network classification model. The model was trained based on the scaled conjugate gradient backpropagation. A nested cross validation with an exhaustive outer validation loop was performed to validate the classification's generalization performance. A high testing prediction accuracy of 97.9% with a Cohen's Kappa coefficient of 97.2% has been interestingly obtained. Prone and supine postures were successfully separated in the classification, in contrast to the majority of previous similar works. We found that using the information of body weight distribution along with the shape and edges contributes to a better classification performance and the ability to separate supine and prone postures. The results are satisfactorily promising toward unobtrusively monitoring posture for ulcer prevention. The method can be used in sleep studies, post-surgical procedures, or applications requiring HBLP identification.

Improved Multi-Grating Filtering Demodulation Method Based on Cascading Neural Networks for Fiber Bragg Grating Sensor

In recent decades, fiber Bragg grating (FBG) sensors have proven useful for structural health monitoring. An accurate and low-cost FBG demodulation method is needed to improve the performance of these sensors in structural-monitoring applications. This paper presents an improved method of matched multi-FBG-filtering demodulation that uses two cascading artificial neural networks (ANNs). The first net is used to select the matched-FBG, and the second net is used to demodulate the sensing signal from the FBG sensor. Several algorithms were tested for training the ANNs. The scaled conjugate gradient backpropagation algorithm proves to be the best algorithm for training the first ANN, and the one-step-secant backpropagation algorithm is most suitable for training the second ANN. Errors in the cascading ANNs can be decreased by adjusting the difference in wavelength between the matched FBGs and varying the algorithms used in the ANNs. When the difference in wavelength is 0.2271 nm, the maximum errors returned with test sets using the optimal algorithms are −10.39 pm and −10.11 μ ϵ for wavelength and strain, respectively. The ANNs prove to be generalizable, given in our results.

Prediction of human assessments of dairy odor utilizing a fast gas chromatograph and neural networks

A method based on hedonic tone was developed and applied to evaluate human assessments of odor emitted from dairy operations using a fast gas chromatograph and neural networks. A general pleasantness scale ranging from −11 (extremely unpleasant) to +11 (extremely pleasant) was used to collect human responses. The panelists were able to identify the difference between various samples, and gave individually consistent responses for the same sample. The measurements of a fast gas chromatograph, called the zNose, were trained using Artificial Neural Networks (ANNs) to predict the human assessments. Three ANNs, Levenberg-Marquardt Back-propagation Neural Network (LMBNN), Scaled Conjugate Gradient Back-propagation (CGBNN), and Resilient Back-propagation Neural Network (RPBNN), were applied to connect human assessments and instrument measurements. In separate validation, zNose-LMBNN model showed superiority in four criteria, Mean Square Error (MSE), Correlation Coefficient (R), probability within 10% range to target, and probability within 5% range to target. The optimal model outputs represented human response as high as 67% within the 10% range and 44% within the 5% range of the targets. In addition, the model outputs have a good linear relationship with the targets (R = 0.53).

Blood Pressure Classification Using the Method of the Modular Neural Networks.

In this paper, we present a new model based on modular neural networks (MNN) to classify a patient's blood pressure level (systolic and diastolic pressure and pulse). Tests are performed with the Levenberg-Marquardt (trainlm) and scaled conjugate gradient backpropagation (traincsg) training methods. The modular neural network architecture is formed by three modules. In the first module we consider the diastolic pressure data; in the second module we use details of the systolic pressure; in the third module, pulse data is used and the response integration is performed with the average method. The goal is to design the best MNN architecture for achieving an accurate classification. The results of the model show that MNN presents an excellent classification for blood pressure. The contribution of this work is related to helping the cardiologist in providing a good diagnosis and patient treatment and allows the analysis of the behavior of blood pressure in relation to the corresponding diagnosis, in order to prevent heart disease.

Evaluation of effective needleless electrospinning parameters controlling polyacrylonitrile nanofibers diameter via modeling artificial neural networks

In this study, we aimed to predict the effects of different needleless electrospinning parameters on the diameter of polyacrylonitrile (PAN) nanofibers via artificial neural network method. The various factors, including polymer solution concentration, applied voltage, and spinneret-to-collector distance, were designed to investigate the diameter of PAN nanofibers produced via needleless electrospinning system. Levenberg–Marquardt, Bayesian regularization, and scaled conjugate gradient back-propagation algorithms were used for the analysis. The results indicated that PAN nanofiber diameters had a direct correlation with the polymer solution concentration and applied voltage and an inverse relation with the nozzle-to-collector distance. The Pearson correlation coefficients were significant at <.01% level for all prediction models. The Bayesian regularization mode I achieved the highest regression value (.9944) for the test data set. The regression value was also calculated and the maximum regression value (.9936) was obtained for the Bayesian regularization model I.

Automated Detection and Classification of Microcalcification Clusters with Enhanced Preprocessing and Fractal Analysis

This paper addresses the automated detection of microcalcification clusters from mammogram images by enhanced preprocessing operations on digital mammograms for automated extraction of breast tissue from background, removing artefacts occurring during image registration using X-rays, followed by fractal analysis of suspicious regions. Identification of breast of either left or right and realigning them to a standard position forms a primitive step in preprocessing of mammograms. As the next step in the process, pectoral muscles are separated. Suspicious regions of microcalcifications are identified and are subjected to further analysis of classifying it as benign or malignant. Texture features are representative of its malignancy and fractal analysis was carried out on extracted suspicious regions for its texture features. Principal Component Analysis was carried out to extract optimal features. Ten features were found to be an optimal number of reduced texture features without compromising on classification accuracy. Scaled conjugate Gradient Back propagation network was used for classification using reduced texture features obtained from PCA analysis. By varying hidden layer neurons, accuracy of results achieved by proposed methods is analysed and is calculated to reach maximum accuracy with an optimal level of 15 neurons. Accuracy of 96.3% was achieved with 10 fractal features as input to neural network and 15 hidden layer neurons in neural network designed. The design of architecture is finalised with maximised accuracy for labelling microcalcification clusters as benign or malignant.

Performance Comparison of Neural Network Training Algorithms for Modeling Customer Churn Prediction

Predicting customer churn has become the priority of every telecommunication service provider as the market is becoming more saturated and competitive. This paper presents a comparison of neural network learning algorithms for customer churn prediction. The data set used to train and test the neural network algorithms was provided by one of the leading telecommunication company in Malaysia. The Multilayer Perceptron (MLP) networks are trained using nine (9) types of learning algorithms, which are Levenberg Marquardt backpropagation (trainlm), BFGS Quasi-Newton backpropagation (trainbfg), Conjugate Gradient backpropagation with Fletcher-Reeves Updates (traincgf), Conjugate Gradient backpropagation with Polak-Ribiere Updates (traincgp), Conjugate Gradient backpropagation with Powell-Beale Restarts (traincgb), Scaled Conjugate Gradient backpropagation (trainscg), One Step Secant backpropagation (trainoss), Bayesian Regularization backpropagation (trainbr), and Resilient backpropagation (trainrp). The performance of the Neural Network is measured based on the prediction accuracy of the learning and testing phases. LM learning algorithm is found to be the optimum model of a neural network model consisting of fourteen input units, one hidden node and one output node. The best result of the experiment indicated that this model is able to produce the performance accuracy of 94.82%.

Abstract WP60: Kernel Spectral Regression and Neural Networks Enable Regional Detection of Hemorrhagic Transformation on Multi-Modal MRI for Acute Ischemic Stroke

Purpose: Hyperperfusion detected on arterial spin labeling (ASL) cerebral blood flow (CBF) images acquired after acute ischemic stroke (AIS) onset has been shown to be significantly associated with development of intracerebral hemorrhage. ASL data is often too noisy for accurate voxel-based analysis alone. The purpose of this study was to develop a machine learning model for the voxel-based detection of tissue at risk for hemorrhagic transformation (HT) utilizing multiple MRI modalities. Materials and Methods: The present study utilized clinical MRI and ASL image data acquired from 67 AIS patients shortly after endovascular therapy. A novel regional cuboid sampling framework was developed for machine learning training, in which local cuboids were extracted from the CBF map, DWI, FLAIR, and TSE images before being matched with GRE-based manually drawn bleed groundtruth delineations. Kernel spectral regression (KSR) uses the eigenvectors of the graph representation to reveal low dimensional structure in high dimensional data. A two-layer feed-forward neural network was then built with 10 neurons in the sigmoid hidden layer and trained with scaled conjugate gradient backpropagation to classify cuboid inputs into likelihood of HT. Results: The proposed multimodal regional framework reached an accuracy of 80.59 ± 3% in detecting hemorrhage with KSR on our dataset (better than a voxel-based prediction on CBF with 72.80 ± 5% accuracy). Using the neural network training, the framework reached an improved accuracy of 95.1% ± 0.6%. Figure 1 shows how the regions with high likelihood of hemorrhage match well with the manually drawn regions in the reference GRE map. Conclusion: Kernel spectral regression and neural networks can provide more accurate detection of tissues at risk for HT. Although CBF can inform AIS patient clinical outcome, the addition of multi-modal MRI data into the regional cuboid framework substantially improves the voxel-based HT detection accuracy.

Analysis of enhanced complex SVR interpolation and SCG-based neural networks for LTE downlink system

In this article, we operate and evaluate the performance of radial basis function (RBF)-based support vector machine regression (SVR) and scaled conjugate gradient back propagation (SCG)-based artificial neural network (ANN), to estimate the channel deviations in frequency domain using the standardised pilot symbols structure for LTE downlink system. We apply complex SVR and ANN to estimate the real vehicular a channel environment well-defined by the International Telecommunications Union (ITU). The suggested procedures use data obtained from the received pilot symbols to estimate the overall frequency response of the frequency selective multipath fading channel in two stages. In the first stage, each technique learns to adjust to the channel fluctuations, then, in the second stage, it predicts all the channel frequency responses. Lastly, in order to assess the abilities of the considered channel estimators, we deliver performance of complex SVR and ANN, which are compared to traditional least squares (LS) and decision feedback (DF) methods. Computer simulation results demonstrate that the complex RBF-based SVR approach has a better precision than other estimation methods.

Analysis of enhanced complex SVR interpolation and SCG-based neural networks for LTE downlink system

In this article, we operate and evaluate the performance of radial basis function (RBF)-based support vector machine regression (SVR) and scaled conjugate gradient back propagation (SCG)-based artificial neural network (ANN), to estimate the channel deviations in frequency domain using the standardised pilot symbols structure for LTE downlink system. We apply complex SVR and ANN to estimate the real vehicular a channel environment well-defined by the International Telecommunications Union (ITU). The suggested procedures use data obtained from the received pilot symbols to estimate the overall frequency response of the frequency selective multipath fading channel in two stages. In the first stage, each technique learns to adjust to the channel fluctuations, then, in the second stage, it predicts all the channel frequency responses. Lastly, in order to assess the abilities of the considered channel estimators, we deliver performance of complex SVR and ANN, which are compared to traditional least squares (LS) and decision feedback (DF) methods. Computer simulation results demonstrate that the complex RBF-based SVR approach has a better precision than other estimation methods.

Prediction of Human Responses to Dairy Odor Using an Electronic Nose and Neural Networks

Abstract. Odor emitted from dairy operations may cause negative reactions by farm neighbors. Identification and evaluation of such malodors is vital for better understanding of human response and methods for mitigating effects of odors. The human nose is a valuable tool for odor assessment, but using human panels can be costly and time-consuming, and human evaluation of odor is subjective. Sensing devices, such as an electronic nose, have been widely used to measure volatile emissions from different materials. The challenge, though, is connecting human assessment of odors with the quantitative measurements from instruments. In this work, a prediction system was designed and developed to use instruments to predict human assessment of odors from common dairy operations. The model targets are the human responses to odor samples evaluated using a general pleasantness scale ranging from -11 (extremely unpleasant) to +11 (extremely pleasant). The model inputs were the electronic nose measurements. Three different neural networks, a Levenberg-Marquardt back-propagation neural network (LMBNN), a scaled conjugate gradient back-propagation neural network (CGBNN), and a resilient back-propagation neural network (RPBNN), were applied to connect these two sources of information (human assessments and instrument measurements). The results showed that the LMBNN model can predict human assessments with accuracy as high as 78% within a 10% range and as high as 63% within a 5% range of the targets in independent validation. In addition, the LMBNN model performed with the best stability in both training and independent validation. Keywords: Animal production, Hedonic tone, Olfactometric models.

Artificial neural network analysis the pulsating Nusselt number and friction factor of TiO2/water nanofluids in the spirally coiled tube with magnetic field

The application of artificial neural network to analyze the pulsating nanofluids heat transfer and pressure drop in the spirally coiled tube with magnetic field are presented. Four different training algorithms of Levenberg-Marquardt Backwardpropagation (LMB), Scaled Conjugate Gradient Backpropagation (SCGB), Bayesian Regulation Backpropagation (BRB), and Resilient Backpropagation (RB) are applied to adjust errors for obtaining the optimal ANN model. The results obtained from the artificial neural network are compared those from the present experiment. It is found that the Levenberg- Marquardt Backpropagation algorithm gives the minimum MSE, and maximum R as compared with other training algorithms. Based on the optimal ANN model, the majority of the data falls within ±2.5%, ±5% of the Nusselt number and friction factor, respectively. The obtained optimal ANN has been applied to predict the thermal performance of the spirally coiled tube with magnetic field.

Scaled conjugate gradient back-propagation algorithm for selection of industrial robots

Purpose: Development of industrial robots and its usage by different manufacturing industries in lot many diverse applications is a very much serious task for the selection of robots. As a consequence, the selection process of the robot becomes incredibly complicated for the potential users because they have a large set of attributes and parameters of the robots are available at their disposal. Moreover, the proposed work gives the efficient decision for the selection of the robots to the potential users of the robots or robot manufacturer. Design/methodology/approach: In this paper, it has been proposed that the scaled conjugate gradient fastest back-propagation algorithm used for the optimized way of selection of robot based on the appropriate parameters. Findings: The ranks of the desired industrial robots are evaluated using the proposed algorithm from the entire best possible robot that specifies the most genuine yardstick of robot selection for the particular application. The complete design performance of the proposed method discussed with different prediction errors such as mean square error, root mean square error, and R-squared error. Originality/value: The proposed methodology is purely original scientific work and very fastest algorithm for the selection of industrial robot. In this work maximum, ten numbers of parameters of the robot are used for the selection of the industrial robot and eight broad categories of the robot are proposed called robot-rank.

Building the Structure and the Neuroemulator Angular Velocity's Learning Algorithm Selection of the Electric Drive of TVR-IM Type

Today, one of the most common ways to control smooth starting and stopping of the induction motors are soft-start system. To ensure such control method the use of closed-speed asynchronous electric drive of TVR-IM type is required. Using real speed sensors is undesirable due to a number of inconveniences exploitation of the drive. The use of the observer based on a neural network is more convenient than the use of the real sensors. Its advantages are robustness, high generalizing properties, lack of requirements to the motor parameters, the relative ease of creation. This article presents the research and selection of the best learning algorithm of the neuroemulator angular velocity of the electric drive of TVR-IM type. There were investigated such learning algorithms as gradient descent back propagation, gradient descent with momentum back propagation, algorithm of Levenberg – Marquardt, scaled conjugate gradient back propagation (SCG). The input parameters of the neuroemulator were the pre treatment signals from the real sensors the stator current and the stator voltage and their delay, as well as a feedback signal from the estimated speed with delay. A comparative analysis of learning algorithms was performed on a simulation model of asynchronous electric drive implemented in software MATLAB Simulink, when the electric drive was running in dynamic mode. The simulation results demonstrate that the best method of learning is algorithm of Levenberg – Marquardt.

Artificial Neural Networks for the Prediction of Compressive Strength of Concrete

In the paper, an artificial neural network (ANN) model is proposed to predict the compressive strength of concrete. For developing the ANN model the data bank on concrete compressive strength has been taken from the experiments conducted in the laboratory under standard conditions. The data set is of two types; in one dataset 15% cement is replaced with fly ash and the other one is without any replacement. Several training algorithms, like Quasi-Newton algorithm with Broyden, Fletcher, Goldfarb, and Shanno (BFGS) update (BFG), Fletcher-reeves conjugate gradient algorithm (CGF), Polak-Ribiere conjugate gradient algorithm (CGP),Powell-Beale conjugate gradient algorithm (CGB), Levenberg–Marquardt (LM), Resilient backpropagation (RP), Scaled conjugate gradient backpropagation (SCG), One step Secant backpropagation (OSS) along with various network architectural parameters are experimentally investigated to arrive at the most suitable model for predicting the compressive strength of concrete. It is found that Levenberg- Marquardt (LM) with tan-sigmoid activation function is best for the prediction of compressive strength of concrete. In-situ concrete compressive strength data, based on varying mix proportions, have been taken from one of the research paper present in literature for the validation of the model. It is also recommended that ANN model with the training function, Levenberg- Marquardt (LM) for the prediction of compressive strength of concrete is one of the best possible tool for the purpose.

Evaluating the performance of artificial neural networks for the classification of freshwater benthic macroinvertebrates

Macroinvertebrates form an important functional component of aquatic ecosystems. Their ability to indicate various types of anthropogenic stressors is widely recognized which has made them an integral component of freshwater biomonitoring. The use of macroinvertebrates in biomonitoring is dependent on manual taxa identification which is currently a time-consuming and cost-intensive process conducted by highly trained taxonomical experts. Automated taxa identification of macroinvertebrates is a relatively recent research development. Previous studies have displayed great potential for solutions to this demanding data mining application. In this research we have a collection of 1350 images from eight different macroinvertebrate taxa and the aim is to examine the suitability of artificial neural networks (ANNs) for automated taxa identification of macroinvertebrates. More specifically, the focus is drawn on different training algorithms of Multi-Layer Perceptron (MLP), probabilistic neural network (PNN) and Radial Basis Function network (RBFN). We performed thorough experimental tests and we tested altogether 13 training algorithms for MLPs. The best classification accuracy of MLPs, 95.3%, was obtained by two conjugate gradient backpropagation variations and scaled conjugate gradient backpropagation. For PNN 92.8% and for RBFN 95.7% accuracies were achieved. The results show how important a proper choice of ANN is in order to obtain high accuracy in the automated taxa identification of macroinvertebrates and the obtained model can outperform the level of identification which is made by a taxonomist.

Predicting expected progeny difference for marbling score in Angus cattle using artificial neural networks and Bayesian regression models

BackgroundArtificial neural networks (ANN) mimic the function of the human brain and are capable of performing massively parallel computations for data processing and knowledge representation. ANN can capture nonlinear relationships between predictors and responses and can adaptively learn complex functional forms, in particular, for situations where conventional regression models are ineffective. In a previous study, ANN with Bayesian regularization outperformed a benchmark linear model when predicting milk yield in dairy cattle or grain yield of wheat. Although breeding values rely on the assumption of additive inheritance, the predictive capabilities of ANN are of interest from the perspective of their potential to increase the accuracy of prediction of molecular breeding values used for genomic selection. This motivated the present study, in which the aim was to investigate the accuracy of ANN when predicting the expected progeny difference (EPD) of marbling score in Angus cattle. Various ANN architectures were explored, which involved two training algorithms, two types of activation functions, and from 1 to 4 neurons in hidden layers. For comparison, BayesCπ models were used to select a subset of optimal markers (referred to as feature selection), under the assumption of additive inheritance, and then the marker effects were estimated using BayesCπ with π set equal to zero. This procedure is referred to as BayesCpC and was implemented on a high-throughput computing cluster.ResultsThe ANN with Bayesian regularization method performed equally well for prediction of EPD as BayesCpC, based on prediction accuracy and sum of squared errors. With the 3K-SNP panel, for example, prediction accuracy was 0.776 using BayesCpC, and ranged from 0.776 to 0.807 using BRANN. With the selected 700-SNP panel, prediction accuracy was 0.863 for BayesCpC and ranged from 0.842 to 0.858 for BRANN. However, prediction accuracy for the ANN with scaled conjugate gradient back-propagation was lower, ranging from 0.653 to 0.689 with the 3K-SNP panel, and from 0.743 to 0.793 with the selected 700-SNP panel.ConclusionsANN with Bayesian regularization performed as well as linear Bayesian regression models in predicting additive genetic values, supporting the idea that ANN are useful as universal approximators of functions of interest in breeding contexts.

An artificial neural network model for virtual Superpave asphalt mixture design

This study presents an artificial neural network (ANN) model to predict the asphalt mixture volumetrics at Superpave gyration levels. The input data-set needed by the algorithm is composed of gradation of the mix, bulk specific gravity of aggregates, low- and high-performance grade of the binder, binder content of the mix and the target number of gyrations (i.e. Nini, Ndes and Nmax). The proposed ANN model uses a three-layer scaled conjugate gradient back-propagation (feed-forward) network. The ANN was trained using data obtained from numerous roads with a total of 1817 different mix designs. Results revealed that the ANN was able to predict Va within Va (measured) ± 1.0% range 85–93% of the time and within Va (measured) ± 0.5% range 60–70% of the time. Currently with the developed ANN model, Superpave mix design can take approximately between 1.5 and 4.5 days, which corresponds to 3–6 days of savings.