Optimal Number Of Hidden Neurons Research Articles

Hybrid artificial neural network models for bearing capacity evaluation of a strip footing on sand based on Bolton failure criterion

This paper employs the Bolton failure criterion, incorporating strength-dilatancy relationships, to analyze the bearing capacity factor of a strip footing on dense sand. Utilizing finite element limit analysis (FELA) based on the lower and upper bound theorems, the study presents the results as average bound solutions. By using the Bolton model, the b parameter is first calibrated and found that it should be about 3.50 to align the ultimate bearing capacity (qu) from FELA to have a good agreement with that from experimental test results from previous studies. The influence of parameters relevant to the Bolton failure criterion is analysed, showing that an increase in relative density (DR) significantly affects the variation in the bearing capacity factor (Nγ) at higher Q values, while lower Q values inhibit dilatancy due to soil crushing. The width of the strip footing (B) has a decreasing effect on Nγ at higher Q values, and the unit weight (γ) changes minimally impact Nγ within the range of 16–22 kN/m3. Additionally, an increase in the critical state friction angle (ϕcv) consistently increases Nγ, highlighting its direct correlation with soil shear strength. A hybrid artificial neural network (ANN) model integrates machine learning with four optimization algorithms: Imperialist Competitive Algorithm (ICA), Ant Lion Optimization (ALO), Teaching Learning Based Optimization (TLBO), and New Self-Organizing Hierarchical Particle Swarm Optimizer with Jumping Time-Varying Acceleration Coefficients (NHPSO-JTVAC). Comparative rank analysis of hybrid ANN models based on the selection of the optimal number of hidden neurons demonstrates that the ANN-TLBO model excels in predicting the bearing capacity factor, achieving a score of 48. This conclusion is corroborated by an error heatmap matrix, which indicates a minimized percentage of error relative to other hybrid ANN models. Importance analysis identifies particle crushing strength (Q) as the most significant factor influencing the bearing capacity factor (Nγ).

The Effect of Biotic Stress in Plant Species Induced by ‘Candidatus Phytoplasma solani’—An Artificial Neural Network Approach

Infections with phytoplasma present one of the most significant biotic stresses influencing plant health, growth, and production. The phytoplasma ‘Candidatus Phytoplasma solani’ infects a variety of plant species. This pathogen impacts the physiological and morphological characteristics of plants causing stunting, yellowing, leaf curling, and other symptoms that can lead to significant economic losses. The aim of this study was to determine biochemical changes in peony (Paeonia tenuifolia L.), mint (Mentha × piperita L.), and dill (Anethum graveolens L.) induced by ‘Ca. Phytoplasma solani’ in Serbia as well as to predict the impact of the biotic stress using artificial neural network (ANN) modeling. The phylogenetic position of the Serbian ‘Ca. Phytoplasma solani’ strains originated from the tested hosts using 16S rRNA (peony and carrot strains) and plsC (mint and dill strains) sequences indicated by their genetic homogeneity despite the host of origin. Biochemical parameters significantly differed in asymptomatic and symptomatic plants, except for total anthocyanidins contents in dill and the capacity of peony and mint extracts to neutralize superoxide anions and hydroxyl radicals, respectively. Principal Component Analysis (PCA) showed a correlation between different chemical parameters and revealed a clear separation among the samples. Based on the ANN performance, the optimal number of hidden neurons for the calculation of TS, RG, PAL, LP, NBT, •OH, TP, TT, Tflav, Tpro, Tant, DPPH, and Car was nine (using MLP 8-9-13), as it produced high r2 values (1.000 during the training period) and low SOS values. Developing an effective early warning system for the detection of plant diseases in different plant species is critical for improving crop yield and quality.

Optimization of Neuro-controller Application for Maximum Power Point Tracking Photovoltaic Systems Through Shannon’s Information Criteria

Due to the extremely poor efficiency of solar energy, researchers created various maximum power point tracking (MPPT) techniques with the aim of enhancing the effectiveness of photovoltaic (PV) systems. Because of their propensity to address complex problems and their non-linear characteristics, Artificial Neural Network (ANN) algorithms are the most frequently used among these MPPT techniques. Nevertheless, the performance of the ANN-based MPP tracking algorithms is contingent upon various factors, including the choice of activation function, the quantity of hidden neurons, and the training algorithm employed. Shannon’s Information Criteria (SIC) is used to determine the optimal number of hidden neurons within a single hidden layer for the neuro-controller application. In this regard, a two-layer Feed-Forward Neural Network (FFNN) was trained using MATLAB/Simulink software, incorporating varying numbers of hidden neurons. The results indicate that the two-layer FFNN with five hidden neurons has the highest performance, as demonstrated by the lowest Mean Squared Error (MSE) of 4.03 × 10-9, which is statistically significant. The successful incorporation of Gaussian noise in the simulation of an 85 kW PV system demonstrates that the ANN-based MPPT algorithm is both theoretically robust and practically viable and reliable for enhancing the efficiency of solar PV systems in real-world scenarios.

An offbeat bolstered swarm integrated ensemble learning (BSEL) model for heart disease diagnosis and classification

In recent years, cardiovascular disease also known as heart disease has emerged as the top cause of mortality throughout the world. Several different cardiac conditions are included. There are many variables that might increase a person's likelihood of developing heart disease; thus, it is critical to develop reliable, rapid methods of diagnosis and treatment. Therefore, the proposed work aims to develop an effective framework, named as, Bolstered Swarm Integrated Ensemble Learning (BSEL) for heart disease detection. Here, the given medical dataset has been cleaned up, transformed, and normalized using the Linear Interpolation Normalization (LIN) technique. For selecting the best features, the Bolstered-up Beetle Swarm Optimization (BBSO) method is used that removes the irrelevant attributes from the standardized datasets. In addition, the best features are employed in the Weighted Ensemble Classification (WEC) model to establish whether or not a patient has heart disease. The accuracy is improved by using an ensemble classifier with the optimal number of hidden neurons, as determined by the Red Colobuses Monkey Optimization (RCoMO) model. The proposed BSEL prediction model's results are tested and assessed during experiment using a variety of performance measures.

Estimation of Shear Wave Velocity for Shallow Depth Using Artificial Neural Network Technique: A Case Study in Rumaila oil field

In Rumaila oilfield, the lost circulation problem is a challenging issue. The geological and geomechanical properties of subsurface formations have a role in causing a circulation loss. One of the natural features related to mud loss against these formations is an unconformity surface. Inferring surface unconformity needs to be understood how the mechanical properties of rocks are distributed across the entire reservoir. The result is identifying areas of loss for solving most loss problems. Using well logs, a geomechanical model can be constructed to identify the surface unconformity. Shear wave velocity is the most crucial factor in determining mechanical properties. They are not frequently recorded during well logging for time and cost-saving purposes. save time and cost. To overcome this challenge, an ANNs model was developed to estimate the missing Vs data for the Hasa and Aruma groups in the Rumaila oil field for interested wells from the south and north domes (extending from the top of Dammam to the bottom of Hartha). The performance of the new model was tested through calibration. The outcomes showed that measured depth (MD), bulk density (RHOB), and compressional velocity (Vp) are key parameters for creating the ANN model utilizing. This study has proven that the basic systematic equations are accurately anticipate shear velocity (Vs) from conventional well logs. The correlation coefficient (R2) and the root mean square error were 0.956) and0.118, respectively.The optimum number of hidden neurons was 3 neurons). The presented model is closely resemble the measured Vs data when dataset from other well was used to check the accuracy of that predictive model. This study presents an effective, simple, and cost-effective technique, which can be used in the absence of rock tests and DTs.

Multiple ECG segments denoising autoencoder model.

Denoising autoencoder (DAE) with a single hidden layer of neurons can recode a signal, i.e., converting the original signal into a noise-reduced signal. The DAE approach has shown a good performance in denoising bio-signals, like electrocardiograms (ECG). In this paper, we study the effect of correlated, uncorrelated and jittered datasets on the performance of the DAE model. Vectors of multiple concatenated ECG segments of simultaneously recorded Einthoven recordings I, II, III are considered to establish the following dataset cases: (1) correlated, (2) uncorrelated, and (3) jittered. We consider our previous work in finding the optimal number of hidden neurons receiving the input signal with respect to signal quality and computational burden by applying Akaike's information criterion. To evaluate DAE, these datasets are corrupted with six types of noise, namely mix noise (MX), motion artifact noise (MA), electrode movement (EM), baseline wander (BW), Gaussian white noise (GWN) and high-frequency noise (HFN), to simulate real case scenario. Spectral analysis is used to study the effects of noise whose power spectrum may overlap with the power spectrum of the wanted signal on DAE performance. The simulation results show (a) that the number of hidden neurons to denoise multiple correlated ECG is much lower than for jittered signals, (b) QRS-complex based ECG alignment preferable, (c) noises with slightly overlapping power spectrum, like BW and HFN, can be easily removed with sufficient number of neurons, while the noise with completely overlapping spectrum, like GWN, requires a very low-dimensional and thus coarser reduction to recover the signal. The performance of DAE model in terms of signal-to-noise ratio improvement and the required number of hidden neurons can be improved by utilizing the correlation among simultaneous Einthoven I, II, III records.

Aspect-based sentiment analysis for hotel reviews using an improved model of long short-term memory

Advances in information technology have given rise to online hotel reservation options. The user review feature is an important factor during the online booking of hotels. Generally, most online hotel booking service providers provide review and rating features for assessing hotels. However, not all service providers provide rating features or recap reviews for every aspect of the hotel services offered. Therefore, we propose a method to summarise reviews based on multiple aspects, including food, room, service, and location. This method uses long short-term memory (LSTM), together with hidden layers and automation of the optimal number of hidden neurons. The F1-measure value of 75.28% for the best model was based on the fact that (i) the size of the first hidden layer is 1,200 neurons with the tanh activation function, and (ii) the size of the second hidden layer is 600 neurons with the ReLU activation function. The proposed model outperforms the baseline model (also known as standard LSTM) by 10.16%. It is anticipated that the model developed through this study can be accessed by users of online hotel booking services to acquire a review recap on more specific aspects of services offered by hotels

Güve-Alev Optimizasyon Algoritması Kullanarak Pleurotus cornucopiae var. citrinopileatus Mantarı Ekstraksiyon Koşullarının Optimizasyonu

Bu çalışmanın amacı, Pleurotus cornucopiae var. citrinopileatus mantarı ekstraksiyon koşullarının yapay zekâ metotlarıyla optimize edilmesidir. Bu amaç doğrultusunda %0, 30, 60, 90 etanol oranı kullanılarak, 1, 2 ve 4 mg/mL ekstrakt konsantrasyonunda elde edilen ekstraktların, 1,2, 3, 4,2 ve 6 pH koşullarındaki antioksidan aktiviteleri mevcut bir deneysel çalışmadan elde edilmiştir. Ekstraksiyon koşullarının modellenmesi Yapay Sinir Ağları (YSA) ile yapılmış ve Güve-Alev Optimizasyon (GAO) algoritması kullanılarak optimize edilmiştir. En iyi tahmin modelini elde etmek için farklı gizli nöron sayıları denenmiş ve en uygun gizli nöron sayısı 5 olarak bulunmuştur. Elde edilen bu modelin hata karelerinin ortalaması ve ortalama mutlak yüzde hatası tüm veri seti için sırasıyla 1,79 ve %3,24 olarak bulunmuştur. Optimizasyon süreci sonrasında maksimum antioksidan aktivite %56,76 olarak bulunmuştur ve bu sonucu elde etmek için optimum ekstraksiyon parametreleri %66,34 etanol oranı, 4 mg/mL ekstrakt konsantrasyonu ve 2,36 pH olarak belirlenmiştir. Bu çalışma, YSA ve GAO algoritmasının birlikte kullanılması ile ekstraksiyon koşullarının optimizasyonunda zaman, emek ve maliyetverimliliği sağlandığını ortaya koymuştur.

Analysis of different artificial neural networks for Bitcoin price prediction

ABSTRACT Predicting the future price of the currency has always been considered one of the most challenging issues. In this paper, we utilize different artificial neural networks (ANNs), including Feedforwardnet, Fitnet, and Cascade networks, and predict the future price of Bitcoin. This paper discusses how a combination of technical attributes, like price-related and lagged features, as inputs of the neural networks, are used to raise the prediction capabilities that directly impact into the final profitability. For empirical analysis, this paper uses the data of the Bitcoin price for a period of 9 months (1.1.2018 - 30.9.2018) available on http://www.coindesk.com. Using a ten-fold cross-validation method, this paper finds the optimal number of hidden neurons for different train functions in each ANN based on error measures, including mean squared error (MSE), mean absolute error (MAE) and mean absolute percentage error (MAPE). Then, the Bitcoin price is predicted, and results are compared based on the amount of R to find out which ANN leads to a better prediction. Finally, this paper concludes that the Fitnet network with trainlm function and 30 hidden neurons outweighs the others. This paper assesses the models’ performance and how specific setups produce principled and stable predictions for beneficial trading.

Application of an Artificial Neural Network Method for the Prediction of the Tube-Side Fouling Resistance in a Shell-And-Tube Heat Exchanger

The accumulation of undesirable deposits on the heat exchange surface represents a critical issue in industrial heat exchangers. Taking experimental measurements of the fouling is relatively difficult and, often, this method does not lead to precise results. To overcome these problems, in the present study, a new approach based on an Artificial Neural Network (ANN) is used to predict the fouling resistance as a function of specific measurable variables in the phosphoric acid concentration process. These include: the phosphoric acid inlet and outlet temperatures, the steam temperature, the phosphoric acid density, the phosphoric acid volume flow rate circulating in the loop. Some statistical accuracy indices are employed simultaneously to justify the interrelation between these independent variables and the fouling resistance and to select the best training algorithm allowing the determination of the optimal number of hidden neurons. In particular, the BFGS quasi-Newton back-propagation approach is found to be the most performing of the considered training algorithms. Furthermore, the best topology ANN for the shell and tube heat exchanger is obtained with a network consisting of one hidden layer with 13 neurons using a tangent sigmoid transfer function for the hidden and output layers. This model predicts the experimental values of the fouling resistance with AARD% = 0.065, MSE = 2.168 × 10−11, RMSE = 4.656 × 10−6 and r2 = 0.994.

Investigation of robustness of hybrid artificial neural network with artificial bee colony and firefly algorithm in predicting COVID-19 new cases: case study of Iran.

As an ongoing public health menace, the novel coronavirus pandemic has challenged the world. With several mutations and a high transmission rate, the virus is able to infect individuals in an exponential manner. At the same time, Iran is confronted with multiple wave peaks and the health care system is facing a major challenge. In consequence, developing a robust forecasting methodology can assist health authorities for effective planning. In that regard, with the help of Artificial Neural Network-Artificial Bee Colony (ANN-ABC) and Artificial Neural Network- Firefly Algorithm (ANN-FA) as two robust hybrid artificial intelligence-based models, the current study intends to select the optimal model with the maximum accuracy rate. To do so, first a sample of COVID-19 confirmed cases in Iran ranging from 19 February 2020 to 25 July 2021 is compiled. 75% (25%) of total observation is randomly allocated as training (testing) data. Afterwards, an ANN model is trained with Levenberg–Marquardt algorithm. Accordingly, based on R-squared and root-mean-square error criteria, the optimal number of hidden neurons is computed as 17. The proposed ANN model is employed to develop ANN-ABC and ANN-FA models for achieving the maximum accuracy rate. According to ANN-ABC, the R- squared values of the optimal model are 0.9884 and 0.9885 at train and test stages. In respect to ANN-FA, the R-squared ranged from 0.9954 to 0.9940 at the train and test phases, which indicates the outperformance of ANN-FA for predicting COVID-19 new cases in Iran. Finally, the proposed ANN-ABC and ANN-FA are applied for simulating the COVID-19 new cases data in different countries. The results revealed that both models can be used as a robust predictor of COVID-19 data and in a majority of cases ANN-FA outperforms the ANN-ABC.

Examination of the ECG signal classification technique DEA-ELM using deep convolutional neural network features

The accurate separation of ECG signals has become crucial to identify heart diseases. Machine learning methods are widely used to separate ECG signals. The aim of this study was to obtain optimal number of hidden neurons of the Extreme Learning Machine (ELM) using the differential evolution algorithm (DEA) and increase the accuracy rate of ECG classification. In this study, a public database on PhysioNet was used for ECG signal classification. A deep feature method using convolutional neural network was used to extract the major features of the ECG samples. Then, a conventional ELM was applied to the ECG signals. Subsequently, the ECG signals with deep properties were shared with the MATLAB classifier toolbox (k-NN, SVM, Decision Trees). In addition, the ECG signals in the dataset were tested using the Genetic Algorithm Wavelet Kernel-ELM (GAWK-ELM). Finally, the DEA-ELM was improved for the determination of the number of hidden neurons. This study optimized the hidden neuron numbers of traditional ELM with DEA using deep learning capabilities in the feature extraction. The aim of was to maximize the best cost of the DEA and achieve the optimal number of hidden neurons in ELM. Accuracy (Acc), sensitivity (Se), specificity (Spe) and F-measure were used as the performance metrics for the classifier performances. The classification results were 80.60%, 81.50%, and 83.12% with SVM, ELM and DEA-ELM, respectively. Thus, the best classification scores were accomplished with an accuracy of 83.12% with the algorithm supported by the DEA.

Stochastic One-Step Training for Feedforward Artificial Neural Networks

This paper studies the use and application of a fast method (non-iterative and instantaneous) for Feedforward Neural Networks training in which the weights of the hidden layer are assigned randomly, and the weights of the output layer are trained through a linear regression adjustment. The method solves two of the problems that are present in traditional training: training time and optimal structure. While traditional iterative training methods require long periods to train a single structure, the proposed method allows training a structure in a single step (not iterative). In this way, by scanning the number of neurons in the hidden layer, many structures are trained in a short time, and it is possible to obtain an optimal topology. A quality control criterion of the predictions is proposed based on the coefficient of determination that guarantees short times and an optimal number of hidden neurons to characterize a specific problem. The feasibility of the proposed method is tested by comparing its performance against building functions of the artificial neural networks toolbox in Matlab®, resulting superior in both approximation quality and training time. A rigorous study and analysis are performed for the regression of simulated data on two different surfaces with a specific noise and different topologies of the neural network. The resulting process time is at least 150 times shorter for proposed training than with the iterative training that Matlab uses, thus obtaining well-founded learning rules. A novel way of an amputated matrix is proposed that breaks the paradigm of the way multiple-output systems are trained and improves the quality of predictions with no detriment to training times.

An improved optimal trigonometric ELM algorithm for numerical solution to ruin probability of Erlang(2) risk model

In this paper, we focus on accurately calculating the numerical solution of the integral-differential equation for ruin probability in Erlang(2) renewal risk model with arbitrary claim distribution. Because the analytical solutions of the equation do not usually exist, firstly, using machine learning method in modern artificial intelligence, the activation functions in the ELM model are changed to trigonometric function, the initial conditions in the integral-differential equation are added to the ELM linear solver to get the ITELM model, and the steps and feasibility of the algorithm are strictly deduced in theory. As the analytic solution for the integral-differential equation only exists when the claim is subject to exponential distribution, and the numerical solution can be gotten with the pareto distribution. And, since the number of hidden neurons in the ITELM model is uncertain, a good numerical value of hidden neurons can only be determined through a large number of iterative tests and comparisons in the actual calculation. Then, we construct a multi-objective optimization model and algorithm, which can get the optimal number of hidden neurons to obtain the IOTELM model and algorithm. Then, in the above two cases for exponential distribution and pareto distribution, the optimal number of hidden neurons is calculated by IOTELM model and algorithm, and then corresponding ITELM models and algorithms are constructed to calculate the corresponding ruin probability. Compared with the previous numerical experiments, it can be seen that the numerical accuracy is greatly improved, which verified the versatility, feasibility and superiority of the proposed IOTELM model and algorithm.

Stability prediction of Himalayan residual soil slope using artificial neural network

In the past decade, advances in machine learning (ML) techniques have resulted in developing sophisticated models that are capable of modelling extremely complex multi-factorial problems like slope stability analysis. The literature review indicates that considerable works have been done in slope stability using ML, but none of them covers the analysis of residual soil slope. The present study aims to develop an artificial neural network (ANN) model that can be employed for evaluating the factor of safety of Shiwalik Slopes in the Himalayan Region. Data obtained from numerical analysis of a residual soil slope were used to develop two ANN models (ANN1 and ANN2 utilising eleven input parameters, and scaled-down number of parameters based on correlation coefficient, respectively). A four-layer, feed-forward back-propagation neural network having the optimum number of hidden neurons is developed based on trial-and-error method. The results derived from ANN models were compared with those achieved from numerical analysis. Additionally, several performance indices such as coefficient of determination (R2), root mean square error, variance account for, and residual error were employed to evaluate the predictive performance of the developed ANN models. Both the ANN models have shown good prediction performance; however, the overall performance of the ANN2 model is better than the ANN1 model. It is concluded that the ANN models are reliable, valid, and straightforward computational tools that can be employed for slope stability analysis during the preliminary stage of designing infrastructure projects in residual soil slope.

Surrogate Model with a Deep Neural Network to Evaluate Gas–Liquid Flow in a Horizontal Pipe

This study developed a data-driven surrogate model based on a deep neural network (DNN) to evaluate gas–liquid multiphase flow occurring in horizontal pipes. It estimated the liquid holdup and pressure gradient under a slip condition and different flow patterns, i.e., slug, annular, stratified flow, etc. The inputs of the surrogate modelling were related to the fluid properties and the dynamic data, e.g., superficial velocities at the inlet, while the outputs were the liquid holdup and pressure gradient observed at the outlet. The case study determined the optimal number of hidden neurons by considering the processing time and the validation error. A total of 350 experimental data were used: 279 for supervised training, 31 for validating the training performance, and 40 unknown data, not used in training and validation, were examined to forecast the liquid holdup and pressure gradient. The liquid holdups were estimated within less than 8.08% of the mean absolute percentage error, while the error of the pressure gradient was 23.76%. The R2 values confirmed the reliability of the developed model, showing 0.89 for liquid holdups and 0.98 for pressure gradients. The DNN-based surrogate model can be applicable to estimate liquid holdup and pressure gradients in a more realistic manner with a small amount of computating resources.

The Yield Prediction of Synthetic Fuel Production from Pyrolysis of Plastic Waste by Levenberg-Marquardt Approach in Feedforward Neural Networks Model.

The conversion of plastic waste into fuel by pyrolysis has been recognized as a potential strategy for commercialization. The amount of plastic waste is basically different for each country which normally refers to non-recycled plastics data; consequently, the production target will also be different. This study attempted to build a model to predict fuel production from different non-recycled plastics data. The predictive model was developed via Levenberg-Marquardt approach in feed-forward neural networks model. The optimal number of hidden neurons was selected based on the lowest total of the mean square error. The proposed model was evaluated using the statistical analysis and graphical presentation for its accuracy and reliability. The results showed that the model was capable to predict product yields from pyrolysis of non-recycled plastics with high accuracy and the output values were strongly correlated with the values in literature.

Analysing machine learning techniques for predicting the hole-filling in pin-in-paste technology

Pin-in-paste technology is an advanced and constantly developing assembly method, where both the surface mounted and through-hole components are joined commonly with the same reflow soldering step. In the paper, detailed evaluation of machine learning based prediction methods was performed, including artificial neural networks (ANN), adaptive neuro fuzzy inference systems (AFNIS) and gradient boosted decision trees to optimize the process parameters of pin-in-paste. The methods were presented for predicting the hole-filling of solder paste during stencil printing and to enable an evaluation of the aforementioned methods for the specific problem. Experiments were performed to obtain the required input data (hole-filling against the process parameters) for the training of the prediction tools. A testboard with plated through-holes of different diameters (0.8, 1, 1.1, 1.4 mm) was applied. Type 4 lead-free solder paste (particle size 20–38 µm) was deposited with stencil printing into the through-holes with different printing speeds (between 20 mm/s and 70 mm/s). The extent of hole-filling was investigated with X-ray analysis. The optimal structure for the prediction method was determined for every approach by varying the size and configuration iteration-wise. Optimal number of hidden neurons was 34 for the full data set and 18 for the smaller – incomplete- test cases. The optimal structure for the ANFIS consists of 8 and 6 membership functions respectively. Triangular and Gaussian membership function types and 7 different training methods were assessed. The ANN, which was trained with Bayesien Regularization (prediction error <15%) was found to be the recommended one for the PIP technology. A sensitivity analysis was carried out serving as a basis for managerial implications, which were also described in the paper. The obtained results aid in improving the quality and reliability of products assembled with pin-in-paste technology, also enable the more precise control in the wake of Industry 4.0 recommendations.

Knowledge Extraction of Adaptive Structural Learning of Deep Belief Network for Medical Examination Data

Deep learning has a hierarchical network structure to represent multiple features of input data. The adaptive structural learning method of Deep Belief Network (DBN) can reach the high classification capability while searching the optimal network structure during the training. The method can find the optimal number of hidden neurons for given input data in a Restricted Boltzmann Machine (RBM) by neuron generation–annihilation algorithm, and generate a new hidden layer in DBN by the extension of the algorithm. In this paper, the proposed adaptive structural learning of DBN (Adaptive DBN) was applied to the comprehensive medical examination data for cancer prediction. The developed prediction system showed higher classification accuracy for test data (99.5% for the lung cancer and 94.3% for the stomach cancer) than the several learning methods such as traditional RBM, DBN, Non-Linear Support Vector Machine (SVM), and Convolutional Neural Network (CNN). Moreover, the explicit knowledge that makes the inference process of the trained DBN is required in deep learning. The binary patterns of activated neurons for given input in RBM and the hierarchical structure of DBN can represent the relation between input and output signals. These binary patterns were classified by C4.5 for knowledge extraction. Although the extracted knowledge showed slightly lower classification accuracy than the trained DBN network, it was able to improve inference speed by about 1/40. We report that the extracted IF-THEN rules from the trained DBN for medical examination data showed some interesting features related to initial condition of cancer.

Artificial neural network modeling-coupled genetic algorithm optimization of supercritical methanol transesterification of Aegle marmelos oil to biodiesel

This study aims to enhance the transesterification process to convert supercritical methanol (SCM) to biodiesel by applying artificial neural network-coupled genetic algorithm (ANN-GA) optimization. The process parameters selected for ANN-GA optimization were temperature of 230–350 °C, methanol-to-oil molar ratio of 24:1 to 48:1, and time of 5–25 min. The initial set of experiments were designed using the central composite design for ANN modeling. The best ANN topology with an optimal number of hidden neurons was obtained by heuristic analysis of coefficient of determination (R) and mean square error values. The R values (1, 0.99986, 0.9998 and 0.99997) for training, validation, testing, and overall prove the precision of the ANN model. Moreover, hypothesis tests were accomplished to enumerate model fitness. The GA-optimized process conditions for SCM transesterification using ANN as the fitness function were temperature of 325.5 °C, methanol-to-oil molar ratio of 41:1, and time of 23 min. Additionally, relative importance analysis was executed to ascertain the most influential variable on the SCM process. The major fuel properties of Aegle marmelos biodiesel were analyzed and found to comply well with American Society for Testing and Materials (ASTM) D6751 standards.