Optimal Multilayer Perceptron Research Articles

Estimation of porosity and volume of shale using artificial intelligence, case study of Kashafrud Gas Reservoir, NE Iran

The key problem in oil exploration and engineering is the lack of accurate and reliable data about the reservoir parameters of a field. Having a precise assessment of petrophysical properties can provide the ability to make decisions with a high degree of confidence about planning for production, exploitation, and further field development scenario. In this research, an artificial intelligence (AI)-based approach was developed to improve the estimation of reservoir parameters including porosity and volume of shale, which has a significant role in different stages of hydrocarbon exploration, in the Kashafrud Gas Reservoir in the northeast of Iran. For this purpose, we measured the petrophysical properties of 27 samples of the Kashafrud Formation. To increase the amount of data for employing a multilayer perceptron (MLP) artificial neural network (ANN), a geostatistical algorithm was used to increase the amount of laboratory measured data of porosity and volume of shale to 686 and 702, respectively. In addition, 2263 well-logging data from the same well were provided. The optimal MLP network with the topology of 6-7-1, and 6-8-1 was selected to estimate the porosity and shale volume with mean squared error (MSE) of 2.78731E−4, and 1.28701E−9, respectively. The training process was performed using two different sets of input data. In the first approach, all available well-logging data were used as input, ending up in high MSE. In the second approach, some selected well logs were used based on the results of sensitivity analysis which clearly improved the estimations. The ability of MLP networks made great improvements in the estimation of the both parameters up to 99.9%. The presence of valuable core data in this study significantly improved the process of comparison and conclusion. The final results prove that AI is a trusted method, also the potential of the ANN method for the reservoir characterization and evaluation associated problems should be taken into consideration. Due to the unavailability of core data along the whole wells, the application of intelligent methods, such as machine learning (ML) can be used to estimate the parameters in other oil or gas fields and wells.

An Indoor Tags Position Perception Method Based on GWO–MLP Algorithm for RFID Robot

This paper proposes a tag position perception method for scenarios such as package retrieval in unmanned warehouses and book management in libraries. This method can accurately predict the distribution of tag space positions in real–time during RFID robot inventory. Firstly, the signal strength (RSSI) and speed of identification (SoI) are used as features. The grey wolf optimization multi–layer perceptron neural network model (GWO–MLP) is employed to predict the distance of tag groups. Secondly, a tag orientation prediction algorithm is designed to estimate the orientation of the tag groups. Finally, the periodicity of the phase is determined by the characteristic of RSSI attenuation as the tag–to–antenna distance increases, solving the problem of position ambiguity caused by phase periodicity. The experiment has shown that this method achieves a high accuracy rate of 96.67% and 97% in predicting the distance and orientation of tag groups, respectively. The average error in distance perception for the single tag is less than 3 cm, enabling precise perception of RFID tag positions. This method facilitates more efficient operation management and accurate item traceability.

A Real-Time Dynamic Gesture Variability Recognition Method Based on Convolutional Neural Networks

Among the many problems in machine learning, the most critical ones involve improving the categorical response prediction rate based on extracted features. In spite of this, it is noted that most of the time from the entire cycle of multi-class machine modeling for sign language recognition tasks is spent on data preparation, including collection, filtering, analysis, and visualization of data. To find the optimal solution for the above-mentioned problem, this paper proposes a methodology for automatically collecting the spatiotemporal features of gestures by calculating the coordinates of the found area of the pose and hand, normalizing them, and constructing an optimal multilayer perceptron for multiclass classification. By extracting and analyzing spatiotemporal data, the proposed method makes it possible to identify not only static features, but also the spatial (for gestures that touch the face and head) and dynamic features of gestures, which leads to an increase in the accuracy of gesture recognition. This classification was also carried out according to the form of the gesture demonstration to optimally extract the characteristics of gestures (display ability of all connection points), which also led to an increase in the accuracy of gesture recognition for certain classes to the value of 0.96. This method was tested using the well-known Ankara University Turkish Sign Language Dataset and the Dataset for Argentinian Sign Language to validate the experiment, which proved effective with a recognition accuracy of 0.98.

Analysis of hydraulic performance in a structured packing column for air/water system: RSM and ANN modeling

The study investigated the hydraulic characteristics of a structured packing column for an air/water system using two modeling approaches: artificial neural network (ANN) and response surface methodology (RSM). Seven parameters were identified as influential factors for pressure drop and liquid hold-up. Two correlations for the responses were proposed using the two-factor interaction (2FI) model based on the RSM-central composite design approach. In the ANN modeling, two algorithms were applied: multi-layer perceptron (MLP) and radial basis function (RBF). The optimal MLP structure using the Bayesian Regularization (trainbr) algorithm had two hidden layers, 16 and 8 neurons, and the optimal RBF structure had 42 neurons. The best MSE performance of the MLP network was 0.0938 and 0.0083 for pressure drop and hold-up, respectively, at 16 epochs. The best MSE performance of the RBF network was 0.0685 and 0.0140 for pressure drop and hold-up, respectively, at 42 epochs. Both the ANN and RSM models provided reliable predictions of the experimental data. In addition, ANN's prediction performance (R2 = 0.993) was somewhat better than RSM's (R2 = 0.926). The results showed that gas factor and liquid load had the greatest effect on pressure drop and liquid hold-up, respectively.

Enhancing multilayer perceptron neural network using archive-based harris hawks optimizer to predict gold prices

The success of the Multi-Layer Perceptron Neural Network (MLP) relies on carefully configuring its weights and biases to promising values. The gradient descent technique is usually used to select the optimal MLP configuration. Such a technique can stuck in local optima and its convergence towards promising regions in the search space is slow. Harris Hawks optimizer (HHO) is a recently proposed metaheuristic technique that mimics the harris’ hawks behavior to chase and catch prey. In this paper, the algorithm search process is enhanced by introducing an external archive that saves the best solutions to be used in the next iterations. Gold price prediction is essential for future investment planning and decision-making in mining projects. The gold price is predicted using a multilayer perceptron (MLP) neural network (NN) that has the optimal control parameters obtained using the improved version of Harris Hawks optimizer (HHO) with an external archive which is called (AHHO-NN). The input features with a complicated relationship can influence gold price prediction models such as inflation rate, precious metals prices, and other features. In this sense, the literature suggests a specific reduced number of features for the MLP-NN model to yield high-quality prediction results. In addition to the prediction model, two feature reduction methods are used to select the proper list of the input features for the MLP-NN model to forecast gold price namely, Pearson’s correlation and a newly proposed categorized correlation. Surprisingly, new features not mentioned in the literature are discovered and some are discarded. The time series dataset used has been extracted from several sources and pre-processed to fit the proposed model. Furthermore, the prediction results have been evaluated using several measures such as prediction accuracy, Mean square error (MSE), Root mean square error (RMSE), Mean absolute error (MAE), and average absolute relative deviation (AARD). The performance of the proposed AHHO-NN model is compared against four swarm intelligence algorithms (i.e., HHO-NN, JAYA-NN, MFO-NN, PSO-NN) as well as four classical machine learning techniques (i.e., linear regression (LR), MLP Regressor (MLP), RANSAC Regressor (RANSAC), and TheilSen Regressor (TR)). Interestingly, the proposed AHHO-NN is able to comparatively produce high-quality prediction results for gold prices with relatively high accuracy. This proves the viability of the proposed AHHO-NN model as well as the proper selection of the primitive input features.

Breast cancer classification using Deep Q Learning (DQL) and gorilla troops optimization (GTO)

Breast cancer (BC) is a primary reason for death among the female population around the world. Early identification can aid in decreasing the mortality rates associated with this disease all over the world. The application of big data in healthcare not only saves lives but also saves time and money. Hospital records, patient medical information, and medical exam results are big data sources in the healthcare sector. In this research, a big data-based two-class (i.e., Benign or Cancer) BC classification model is developed using the Deep Reinforcement Learning (DRL) method. The model stages are big data collection, preprocessing, feature selection, classification, and explanations. In the preprocessing step, the data is normalized, and the missing values are replaced. The gorilla troops optimization (GTO) algorithm is employed for the feature selections. Deep reinforcement learning-based Deep Q learning (DQL) is used for the classification, and LIME is used to explain the predicted output. Three different datasets such as WBCD, WDBC, and WPBC from the UCI repository, are used for evaluation. We verify the proposed model against the radial basis function ensemble boosting learning method (RBF-ELB), the Particle Swarm Optimization Multilayer Perceptron (PSO-MLP), and the Genetic Algorithm Multilayer Perceptron (GA-MLP). The results of the experiments show that our method outperforms the traditional methods. The proposed GTO-DQL model achieves 98.90% accuracy for the WBCD dataset, 99.02% for WDBC, and 98.88% for the WPBC dataset, respectively.

Prediction model for suicide based on back propagation neural network and multilayer perceptron.

The aim was to explore the neural network prediction model for suicide based on back propagation (BP) and multilayer perceptron, in order to establish the popular, non-invasive, brief and more precise prediction model of suicide. Data were collected by psychological autopsy (PA) in 16 rural counties from three provinces in China. The questionnaire was designed to investigate factors for suicide. Univariate statistical methods were used to preliminary filter factors, and BP neural network and multilayer perceptron were employed to establish the prediction model of suicide. The overall percentage correct of samples was 80.9% in logistic regression model. The total coincidence rate for all samples was 82.9% and the area under ROC curve was about 82.0% in the Back Propagation Neural Network (BPNN) prediction model. The AUC of the optimal multilayer perceptron prediction model was above 90% in multilayer perceptron model. The discrimination efficiency of the multilayer perceptron model was superior to BPNN model. The neural network prediction models have greater accuracy than traditional methods. The multilayer perceptron is the best prediction model of suicide. The neural network prediction model has significance for clinical diagnosis and developing an artificial intelligence (AI) auxiliary clinical system.

Thermo-physical properties and heat transfer potential of novel silica-ethylene glycol mono nanofluid: Experiments and multi-layer perceptron (MLP) modelling

The aim of this research is to estimate the thermophysical properties and heat transfer potential of silica-ethylene glycol (EG) mono nanofluid. Silica nanoparticles were dispersed in pure ethylene glycol in four different concentrations of 0.01, 0.05, 0.10 and 0.50 vol% via two-step method. The dispersion stability of the nanofluid was enhanced including Gum Arabic (GA) as surfactant. Colorimeter and Zeta potential analyses were employed to determine the stability of the as-prepared samples. Thermal conductivity of nanofluid was determined by innovative sonic velocity measurement technique. Viscosity was measured by capillary viscometer. The highest augmentation in thermal conductivity was found to be 19.19% at 60 °C and in viscosity was found to be 5.07% at 40 °C with the 0.50 vol% SiO2 particle dispersion. A set of new data fitting-based correlations and optimal multi-layer perceptron (MLP) models for thermal conductivity and viscosity of Silica-EG nanofluids with high prediction capacities (R-square > 90%) were proposed based on the experimental data. The MLP models were more accurate than fitting models. Evaluation criterion was implemented to estimate the potential heat transfer of nanofluid.

Neutrosophic C-Means Clustering with Optimal Machine Learning Enabled Skin Lesion Segmentation and Classification

Early detection and classification of skin lesions using dermoscopic images have attracted significant attention in the healthcare sector. Automated skin lesion segmentation becomes tedious owing to the presence of artifacts like hair, skin line, etc. Earlier works have developed skin lesion detection models using clustering approaches. The advances in neutrosophic set (NS) models can be applied to derive effective clustering models for skin lesion segmentation. At the same time, artificial intelligence (AI) tools can be developed for the identification and categorization of skin cancer using dermoscopic images. This article introduces a Neutrosophic C-Means Clustering with Optimal Machine Learning Enabled Skin Lesion Segmentation and Classification (NCCOML-SKSC) model. The proposed NCCOML-SKSC model derives a NCC-based segmentation approach to segment the dermoscopic images. Besides, the AlexNet model is exploited to generate a feature vector. In the final stage, the optimal multilayer perceptron (MLP) model is utilized for the classification process in which the MLP parameters are chosen by the use of a whale optimization algorithm (WOA). A detailed experimental analysis of the NCCOML-SKSC model using a benchmark dataset is performed and the results highlighted the supremacy of the NCCOML-SKSC model over the recent approaches.

High-resolution prediction of quenching behavior using machine learning based on optical fiber temperature measurement

Quenching heat transfer is a representative complex phenomenon in thermal-hydraulic engineering. Despite the tremendous efforts to precisely predict the quenching behavior, conventional analysis methodology and correlations have exhibited limited prediction capacity on quenching heat transfer according to axial locations of heater rods. The deviations of existing models result from the uncertainty of axial heat conduction with low-resolution temperature measurement and limited regression performance. In this study, machine learning models were trained with optical fiber temperature measurement data having high spatial resolution about quenching to overcome the intrinsic limitation of conventional prediction methods. Support vector machine (SVM), random forest (RF), and multi-layer perceptron (MLP) models were trained, and the MLP model showed best prediction performance (R2 = 0.9999 and test RMSE = 1.41 °C) due to its strong analysis of the underlying relationship between input and output with interpolation capacity. The optimal MLP model (5-30-30-30-1 architecture) showed good agreement with experimental data for bottom quenching behaviors of Inconel-600 and Monel K500, in aspects of minimum film boiling temperature, quench front propagation velocity, and transient boiling curve by providing spatially resolutive quenching behavior that cannot be obtained from conventional correlations and having high uncertainty in existing computational analysis methods. The developed MLP model has the capability to predict the quenching behavior of FeCrAl accident tolerant fuel cladding surface, and better predictability on minimum film boiling temperature (average error of 3.7%) than conventional correlation having 7.1% average error. The suggested methodology using machine learning technique will contribute to innovatively improve the predictability on quenching phenomena with reducing uncertainty.

A neuro evolutionary algorithm for patient calibrated prediction of survival in Glioblastoma patients

Background and objectivesGlioblastoma multiforme (GBM) is the most common and malignant type of primary brain tumors. Radiation therapy (RT) plus concomitant and adjuvant Temozolomide (TMZ) constitute standard treatment of GBM. Existing models for GBM growth do not consider the effect of different schedules on tumor growth and patient survival. However, clinical trials show that treatment schedule and drug dosage significantly affect patient survival. The goal is to provide a patient calibrated model for predicting survival according to the treatment schedule. MethodsWe propose a top-down method based on artificial neural networks (ANN) and genetic algorithm (GA) to predict survival of GBM patients. A feed forward undercomplete Autoencoder network is integrated with the neuro-evolutionary (NE) algorithm in order to extract a compressed representation of input clinical data. The proposed NE algorithm uses GA to obtain optimal architecture of a multi-layer perceptron (MLP). Taguchi L16 orthogonal design of experiments is used to tune parameters of the proposed NE algorithm. Finally, the optimal MLP is used to predict survival of GBM patients. ResultsData from 8 related clinical trials have been collected and integrated to train the model. From 847 evaluable cases, 719 were used for train and validation and the remaining 128 cases were used to test the model. Mean absolute error of the predictions on the test data is 0.087 months which shows excellent performance of the proposed model in predicting survival of the patients. Also, the results show that the proposed NE algorithm is superior to other existing models in both the mean and variability of the prediction error.

Landslide hazard risk modeling in north-west of Iran using optimized machine learning models

Landslide is one of the most destructive natural hazards threatening the life of people and their properties. Mountainous areas are at a high risk of landslide occurrence. One-third of Iran is mountainous areas where landslide susceptibility modeling is required for sustainable land development. In this research for landslide susceptibility modeling, several supervised machine learning models, including Random Forest, eXtreme Gradient Boosting, the Salp Swarm Algorithm Multi-layer Perceptron (SSAMLP), and the Multi-verse Optimization Multi-layer Perceptron (MVOMLP), are developed in R and MATLAB programming languages. The Random Forest and eXtreme Gradient Boosting models are optimized based on their hyper-parameters in the R programming language using the caret library. In this research, landslide susceptibility models divide the study area into very low (class 1), low (class 2), moderate (class 3), high (class 4), and very high (class 5) landslide susceptibility zones. Results show that the eXtreme Gradient Boosting model has the best performance with values of 0.00225 and 0.00805 for the Mean Absolute Error and Root Mean Squared Error indices, respectively, for the training dataset. For the test dataset, the eXtreme Gradient Boosting model with the values of 0.0599 for the Mean Absolute Error, respectively, outperformed the Random Forest and the MVOMLP and the SSAMLP models. In terms of Mean Absolute Error, the SSAMLP algorithm has the best performance over the other three supervised models with a value of 0.1472.

Detection of liner surface defects in solid rocket motors using multilayer perceptron neural networks

Debonding problems along the propellant/liner/insulation interface are a critical point to the integrity and one of the major causes of structural failures of solid rocket motors. Current solutions are typically restricted to methods for assessing the integrity of the rocket motors structure and visually inspecting their components. In this context, this paper presents an improved algorithm to detect liner surface defects that may compromise the bonding between the solid propellant and the insulation. The use of Local Binary Patterns (LBP) provides a structural and statistical approach to texture analysis of liner sample images. Along with color information extraction, these two methods allow the representation of image pixels by feature vectors that are further processed by a Multilayer Perceptron (MLP) neural network classifier. The MLP neural network analyzes liner sample images and classifies each pixel into one of three classes: non-defect, foreign object, and defect. Several tests were executed varying different parameters to find the optimal MLP configuration, and as a result, the best classification accuracy of 99.08%, 90.66%, and 99.48% was achieved for the corresponding classes. Moreover, the defect size estimate showed that the MLP classifier correctly identified defects less than 1 mm long, with a relatively small number of training examples. Positive results indicate that the algorithm can identify liner surface defects with a performance similar to human inspectors and has the potential to assist or even automate the liner inspection process of solid rocket motors.

Shallow Landslide Susceptibility Models Based on Artificial Neural Networks Considering the Factor Selection Method and Various Non-Linear Activation Functions

Landslide susceptibility mapping is well recognized as an essential element in supporting decision-making activities for preventing and mitigating landslide hazards as it provides information regarding locations where landslides are most likely to occur. The main purpose of this study is to produce a landslide susceptibility map of Mt. Umyeon in Korea using an artificial neural network (ANN) involving the factor selection method and various non-linear activation functions. A total of 151 historical landslide events and 20 predisposing factors consisting of Geographic Information System (GIS)-based morphological, hydrological, geological, and land cover datasets were constructed with a resolution of 5 x 5 m. The collected datasets were applied to information gain ratio analysis to confirm the predictive power and multicollinearity diagnosis to ensure the correlation of independence among the landslide predisposing factors. The best 11 predisposing factors that were selected in this study were randomly divided into a 70:30 ratio for training and validation datasets, which were used to produce ANN-based landslide susceptibility models. The ANN model used in this study had a multi-layer perceptron (MLP) structure consisting of an input layer, one hidden layer, and an output layer. In the output layer, the logistic sigmoid function was used to represent the result value within the range of 0 to 1, and six non-linear activation functions were used for the hidden layer. The performance of the landslide susceptibility models was evaluated using the receiver operating characteristic curve, Kappa index, and five statistical indices (sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV)) with the training dataset. In addition, the landslide susceptibility models were validated using the aforementioned measures with the validation dataset and were compared using the Friedman test to check the significant differences among the six developed models. The optimal number of neurons was determined based on the aforementioned performance evaluation and validation results. Overall, the model with the best performance was the MLP model with the logistic sigmoid activation function in the output layer and the hyperbolic tangent sigmoid activation function with five neurons in the hidden layer. The validation results of the best model showed a sensitivity of 82.61%, specificity of 78.26%, accuracy of 80.43%, PPV of 79.17%, NPV of 81.82%, a Kappa index of 0.609, and AUC of 0.879. The results of this study highlight the effectiveness of selecting an optimal MLP model structure for shallow landslide susceptibility mapping using an appropriate predisposing factor section method.

A hybrid grasshopper and new cat swarm optimization algorithm for feature selection and optimization of multi-layer perceptron

The classification accuracy of a multi-layer perceptron (MLP) depends on the selection of relevant features from the data set, its architecture, connection weights and the transfer functions. Generating an optimal value of all these parameters together is a complex task. Metaheuristic algorithms are popular choice among researchers to solve complex optimization problems. This paper presents a hybrid metaheuristic algorithm simple matching-grasshopper new cat swarm optimization algorithm (SM-GNCSOA) that optimizes all the four components simultaneously. SM-GNCSOA uses grasshopper optimization algorithm, a new variant of binary grasshopper optimization algorithm called simple matching-binary grasshopper optimization algorithm and a new variant of cat swarm optimization algorithm called new cat swarm optimization algorithm to generate an optimal MLP. Features play a vital role in determining the classification accuracy of a classifier. Here, we propose a new feature penalty function and use it in SM-GNCSOA to prevent underfitting or overfitting due to the selected number of features. To evaluate the performance of SM-GNCSOA, different variants of SM-GNCSOA are proposed and their classification accuracies are compared with SM-GNCSOA on ten classification data sets. The results show that SM-GNCSOA gives better results on most of the data sets due to its capability to balance exploration and exploitation and to avoid local minima.

Predicting Inpatient Payments Prior to Lower Extremity Arthroplasty Using Deep Learning: Which Model Architecture Is Best?

BackgroundRecent advances in machine learning have given rise to deep learning, which uses hierarchical layers to build models, offering the ability to advance value-based healthcare by better predicting patient outcomes and costs of a given treatment. The purpose of this study is to compare the performance of 2 common deep learning models, traditional multilayer perceptron (MLP), and the newer dense neural network (DenseNet), in predicting outcomes for primary total hip arthroplasty (THA) and total knee arthroplasty (TKA) as a foundation for future musculoskeletal studies seeking to utilize machine learning. MethodsUsing 295,605 patients undergoing primary THA and TKA from a New York State inpatient administrative database from 2009 to 2016, 2 neural network designs (MLP vs DenseNet) with different model regularization techniques (dropout, batch normalization, and DeCovLoss) were applied to compare model performance on predicting inpatient procedural cost using the area under the receiver operating characteristic curve (AUC). Models were implemented to identify high-cost surgical cases. ResultsDenseNet performed similarly to or better than MLP across the different regularization techniques in predicting procedural costs of THA and TKA. Applying regularization to DenseNet resulted in a significantly higher AUC as compared to DenseNet alone (0.813 vs 0.792, P = .011). When regularization methods were applied to MLP, the AUC was significantly lower than without regularization (0.621 vs 0.791, P = 1.1 × 10−15). When the optimal MLP and DenseNet models were compared in a head-to-head fashion, they performed similarly at cost prediction (P > .999). ConclusionThis study establishes that in predicting costs of lower extremity arthroplasty, DenseNet models improve in performance with regularization, whereas simple neural network models perform significantly worse without regularization. In light of the resource-intensive nature of creating and testing deep learning models for orthopedic surgery, particularly for value-centric procedures such as arthroplasty, this study establishes a set of key technical features that resulted in better prediction of inpatient surgical costs. We demonstrated that regularization is critically important for neural networks in arthroplasty cost prediction and that future studies should utilize these deep learning techniques to predict arthroplasty costs. Level of EvidenceIII.

MLP-LOA: a metaheuristic approach to design an optimal multilayer perceptron

Designing an ANN is a complex task as its performance is highly dependent on the network architecture as well as the training algorithm used to select proper synaptic weights and biases. Choosing an optimal design leads to greater accuracy when the ANN is used for classification. In this paper, we propose an approach multilayer perceptron-lion optimization algorithm (MLP-LOA) that uses lion optimization algorithm to find an optimum multilayer perceptron (MLP) architecture for a given classification problem. MLP-LOA uses back-propagation (BP) for training during the optimization process. MLP-LOA also optimizes learning rate and momentum as they have a significant role while training MLP using BP. LOA is a population-based metaheuristic algorithm inspired by the lifestyle of lions and their cooperative behavior. LOA, unlike other metaheuristics, uses different strategies to search for optimal solution, performs strong local search and helps to escape from worst solutions. A new fitness function is proposed to evaluate MLP based on its generalization ability as well as the network’s complexity. This is done to avoid dense architectures as they increase chances of overfitting. The proposed approach is tested on different classification problems selected from University of California Irvine repository and compared with the existing state-of-the-art techniques in terms of accuracy achieved during testing phase. Experimental results show that MLP-LOA performs better as compared to the existing state-of-the-art techniques.

Prediction of CH4 adsorption on different activated carbons by developing an optimal multilayer perceptron artificial neural network

ABSTRACTA deep understanding of adsorption processes is essential for the design and optimization of industrial units. Storage of methane adsorbed on activated carbon (AC) at low pressure and room temperature (adsorbed natural gas) has been studied in recent years as an alternative model to compressed natural gas and liquefied natural gas technologies. The current study plays a significant role in modeling CH4 adsorption on different ACs through the optimal multilayer perceptron (MLP) neural network . Therefore, lots of adsorption data points were used for modeling. To optimize the efficiency of a predictive model, two optimization algorithms including LevenbergMarquardt (LM) and Bayesian regularization were utilized to find the optimal models’ parameters during prediction analysis. In order to demonstrate the efficiency of the proposed method, it is compared with several other experimental data points. Results of optimizations indicate the superiority of the proposed method over the other techniques, and forecasting error is remarkably reduced. As a result, it was found that the MLP-LM is the more accurate model for estimating CH4 adsorption with root-mean-square error and coefficient of determination of 0.00025 and 0.9921, respectively.

Recognition of human gait in oblique and frontal views using Kinect

This study describes the recognition of human gait in the oblique and frontal views using novel gait features derived from the skeleton joints provided by Kinect. In D-joint, the skeleton joints were extracted directly from the Kinect, which generates the gait feature. On the other hand, H-joint distance is a feature of distance between the hip joint with other skeleton joints. Prior to the gait feature extraction, the skeleton joints provided by Kinect were pre-processed in order to standardize the size of the skeleton image as well as to detect the gait feature within a full gait cycle. To classify gait patterns according to its own group, a multi-layer perceptron was employed in the pattern recognition stage. Results show that a perfect recognition of human gait (100%) was attained for the frontal view using the feature of H-joint distance at the optimal multi-layer perceptron (20 hidden units).

Application of evolutionary neural networks and support vector machines to model NOx emissions from gas turbines

Evolutionary artificial neural networks and support vector machines were investigated for the modeling of NOx emissions from gas turbines. For ANN’s, a genetic algorithm (GA) with indirect binary encoding was employed to obtain optimal multi-layer perceptron (MLP)-type architectures. The contribution of this research includes the use of an improved GA objective function that takes advantage of the number of effective parameters provided by Bayesian regularization. For SVM’s, a GA is also used with a search space covering all hyperparameters, including the choice of the kernel itself. The objective function took advantage of the number of support vectors to yield models that are more capable of generalization and are less computationally expensive. To test the performance of these models, three datasets describing NOx emissions from three different turbines were used. In previous works, these emissions were modeled using a shallow network architecture obtained by trial and error. When the performance of the models was compared, the optimized ANN’s reduced the mean squared error (MSE) over testing sets by up to 90%, and optimized SVM’s led to a reduction of up to 70%; with the latter’s greatest advantage being computational efficiency and consistency with the different GA runs. The GA converged in a period in the order of minutes, which is more efficient than trial-and-error procedures. This work shows that more attention should be given to the influence of machine learning model architectures on the accuracy of PEMS models, as the computational cost of the GA is well justified by the resulting high accuracy.