Neural Network Methodology Research Articles

Evaluation and Exploration of Machine Learning and Convolutional Neural Network Classifiers in Detection of Lung Cancer from Microarray Gene-A Paradigm Shift.

Microarray gene expression-based detection and classification of medical conditions have been prominent in research studies over the past few decades. However, extracting relevant data from the high-volume microarray gene expression with inherent nonlinearity and inseparable noise components raises significant challenges during data classification and disease detection. The dataset used for the research is the Lung Harvard 2 Dataset (LH2) which consists of 150 Adenocarcinoma subjects and 31 Mesothelioma subjects. The paper proposes a two-level strategy involving feature extraction and selection methods before the classification step. The feature extraction step utilizes Short Term Fourier Transform (STFT), and the feature selection step employs Particle Swarm Optimization (PSO) and Harmonic Search (HS) metaheuristic methods. The classifiers employed are Nonlinear Regression, Gaussian Mixture Model, Softmax Discriminant, Naive Bayes, SVM (Linear), SVM (Polynomial), and SVM (RBF). The two-level extracted relevant features are compared with raw data classification results, including Convolutional Neural Network (CNN) methodology. Among the methods, STFT with PSO feature selection and SVM (RBF) classifier produced the highest accuracy of 94.47%.

Predicting the Rheological Properties of Bitumen-Filler Mastic Using Machine Learning Techniques

This study uses the artificial neural network and response surface methodology to develop two models for predicting the rheological properties, complex modulus (G*) and phase angle (δ) of bitumen-filler mastic. It also analyses and evaluates the accuracy of both models by determining the coefficient of determination (R2), mean squared error (MSE), and root mean squared error (RMSE). The prediction models use the G* and δ data from a previous study by researchers at the Nottingham Transportation Engineering Centre to determine three types of bitumen-filler mastic (limestone, cement and grit stone) with varying filler concentrations of 15, 35, 40 and 65%. The analysis shows that both models perform well in predicting the rheological properties of bitumen-filler mastic. A comparison of the two models shows that the artificial neural network (ANN) has higher accuracy than the response surface methodology model, with an R2 value exceeding 0.92. The results of the ANN achieve a higher R2 value and lower MSE and RMSE values. In summary, the performance of the artificial neural network model is better than the response surface methodology model, which uses the full quadratic, pure quadratic, linear and interaction mathematical methods.

Dry-sliding wear properties of 3D printed PETG/SCF/OMMT nanocomposites: Experimentation and model predictions using artificial neural network

In this article, an attempt has been made to experimentally investigate the synergistic effect of organically modified montmorillonite (OMMT) nanoclay and short carbon fibers (SCFs) on the tribological behaviour of additively manufactured Polyethylene Terephthalate Glycol (PETG) based nanocomposites. The tribo-specimens are 3D printed using fused deposition modelling (FDM). The tribological properties, that is, specific wear rate (SWR) and coefficient of friction (CoF) of various PETG/SCF/OMMT nanocomposites were assessed by performing dry-sliding wear test. In addition, an artificial neural network (ANN) methodology is proposed to accurately predict the wear performance of PETG nanocomposites. The ANN model is trained using the datasets obtained from the experimentation. For the training of the ANN model, the Levenberg–Marquardt optimisation algorithm with 10 neurons along with a tangent sigmoid activation function is utilised. Additional experimentation was performed for arbitrary load and sliding velocity which were not used for training the ANN model and the results were compared to assess the predictive capability of the ANN model on unseen data. The proposed ANN methodology predicted the SWR and CoF with agreeable accuracy. It is believed that by adopting the proposed ANN methodology, the experimentation costs and time can be significantly reduced without compromising on the accuracy of the results.

A Physics-Informed Recurrent Neural Network for RRAM Modeling

Extracting behavioral models of RRAM devices is challenging due to their unique “memory” behaviors and rapid developments, for which well-established modeling frameworks and systematic parameter extraction processes are not available. In this work, we propose a physics-informed recurrent neural network (PiRNN) methodology to generate behavioral models of RRAM devices from practical measurement/simulation data. The proposed framework can faithfully capture the evolution of internal state and its impacts on the output. A series of modifications informed by the RRAM device physics are proposed to enhance the modeling capabilities. The integration strategy of Verilog-A equivalent circuits, is also developed for compatibility with existing general-purpose circuit simulators. The Verilog-A model can be easily adopted into the SPICE-type simulator for the circuit design with a variable step that differs from the training process. Numerical experiments with real RRAM devices data demonstrate the feasibility and advantages of the proposed methodology.

Stimulated Raman histology as a method to determine the adequacy of renal mass biopsy and identify malignant subtypes of renal cell carcinoma.

Renal tumor biopsy requires adequate tissue sampling to aid in the investigation of small renal masses. In some centers the contemporary nondiagnostic renal mass biopsy rate may be as high as 22% and may be as high as 42% in challenging cases. Stimulated Raman Histology (SRH) is a novel microscopic technique which has created the possibility for rapid, label-free, high-resolution images of unprocessed tissue which may beviewed onstandard radiology viewing platforms. The application of SRH to renal biopsy may provide the benefits of routine pathologic evaluation during the procedure, thereby reducing nondiagnostic results. We conducted a pilot feasibility study, to assess if renal cell carcinoma (RCC) subtypes may be imaged and to see if high-quality hematoxylin and eosin (H&E) could subsequently be generated. An 18-gauge core needle biopsy was taken from a series of 25 ex vivo radical or partial nephrectomy specimens. Histologic images of the fresh, unstained biopsy samples were obtained using a SRH microscope using 2 Raman shifts: 2,845 cm-1and 2,930 cm-1. The cores were then processed as per routine pathologic protocols. The SRH images and hematoxylin and eosin (H&E) slides were then viewed by a genitourinary pathologist. The SRH microscope took 8 to 11 minutes to produce high-quality images of the renal biopsies. Total of 25 renal tumors including 1 oncocytoma, 3 chromophobe RCC, 16 clear cells RCC, 4 papillary RCC, and 1 medullary RCC were included. All renal tumor subtypes were captured, and the SRH images were easily differentiated from adjacent normal renal parenchyma. High quality H&E slides were produced from each of the renal biopsies after SRH was completed. Immunostains were performed on selected cases and the staining was not affected by the SRH image process. SRH produces high quality images of all renal cell subtypes that can be rapidly produced and easily interpreted to determine renal mass biopsy adequacy, and on occasion, may allow renal tumor subtype identification. Renal biopsies remained available to produce high quality H&E slides and immunostains for confirmation of diagnosis. Procedural application has promise to decrease the known rate of renal mass nondiagnostic biopsies, and application of convolutional neural network methodology may further improve diagnostic capability and increase utilization of renal mass biopsy among urologists.

Facial Skin Type Detection for Race Classification using Convolutional Neural Network and Haar Cascade Method

Every organism possesses a unique structural makeup, extending from molecules to organ systems. One such organ, the skin, is our body's largest and the most complex. Its diversity in color corresponds with various human races, although the facial skin type, an often-overlooked factor, also plays a significant role in race identification. In this research, a system was developed to recognize racial classifications from facial skin types by focusing on features in the facial T and U areas, using the Convolutional Neural Network (CNN) and Haar cascade methodologies. CNN was employed due to its ability to use the convolution process, moving a kernel across an image, multiplying it with the applied filter, and thereby generating new representative information. It is especially effective in image recognition and processing. The Haar cascade method, on the other hand, was used to outline the T and U areas on the face for the skin type detection system. The T area, known for oil detection, identifies skin types, while the U area identifies race types by forming facial patterns. This system, trained on 1670 race and 60 skin type datasets and optimized using the Adam optimizer, exhibited high accuracy levels. Upon testing with five new samples, it demonstrated an average accuracy of 98% in race detection and 97% in skin type detection.

Global attention-enabled texture enhancement network for MR image reconstruction.

Although recent convolutional neural network (CNN) methodologies have shown promising results in fast MR imaging, there is still a desire to explore how they can be used to learn the frequency characteristics of multicontrast images and reconstruct texture details. A global attention-enabled texture enhancement network (GATE-Net) with a frequency-dependent feature extraction module (FDFEM) and convolution-based global attention module (GAM) is proposed to address the highly under-sampling MR image reconstruction problem. First, FDFEM enables GATE-Net to effectively extract high-frequency features from shareable information of multicontrast images to improve the texture details of reconstructed images. Second, GAM with less computation complexity has the receptive field of the entire image, which can fully explore useful shareable information of multi-contrast images and suppress less beneficial shareable information. The ablation studies are conducted to evaluate the effectiveness of the proposed FDFEM and GAM. Experimental results under various acceleration rates and datasets consistently demonstrate the superiority of GATE-Net, in terms of peak signal-to-noise ratio, structural similarity and normalized mean square error. A global attention-enabled texture enhancement network is proposed. it can be applied to multicontrast MR image reconstruction tasks with different acceleration rates and datasets and achieves superior performance in comparison with state-of-the-art methods.

A novel artificial neural network methodology to produce high-resolution bioclimatic maps using Earth Observation data: A case study for Cyprus

The aim of this research is to propose a novel methodology that exploits Earth Observation (EO) data to accurately produce high-resolution bioclimatic maps at large spatiotemporal scales. This method directly links EO products (i.e., land surface temperature - LST and Normalized Difference Vegetation Index - NDVI) to air temperature (Tair) and such thermal indices as the Universal Thermal Climate Index (UTCI), and the Physiologically Equivalent Temperature (PET) to produce large-scale high-quality bioclimatic maps at a spatial resolution of 100 m. The proposed methodology is based on Artificial Neural Networks (ANNs), and the bioclimatic maps are developed with the use of Geographical Information Systems. High-resolution LST maps are produced from the spatial downscaling of EO images and the application of the methodology in the case of the island of Cyprus highlights the ability of EO parameters to estimate accurately Tair as well as the above mentioned thermal indices. The results are validated for different conditions and the overall Mean Absolute Error for each case ranges from 1.9 °C for Tair to 2.8 °C for PET and UTCI. The trained ANNs could be used in near real-time for estimating the spatial distribution of outdoor thermal conditions and for assessing the relationship between human health and the outdoor thermal environment. On the basis of the developed bioclimatic maps, high-risk areas were identified. Furthermore, the study examines the relationship between land cover and Tair, UTCI, and PET, and the results provide evidence of the suitability of the method to monitor the dynamics of the urban environment and the effectiveness of urban nature-based solutions. Studies on bioclimate analysis monitor thermal environment, raise awareness and enhance the capacity of national public health systems to respond to thermally-induced health risks.

Neural Network Analysis as a Base of the Future System of Water–Environmental Regulation

The article considers neural-network methods and technologies, which are relatively new even for many researchers and experts, as applied to water–environmental regulation. The efficiency of neural networks in this line of studies is due to their self-training, and the ensuing ability to reveal complex nonlinear relationships between the characteristics under control by data processing instruments, consisting of interrelated neurons. The methodology of artificial neural networks and the features of their functioning are described. Training and methodological examples are given to illustrate their potential use. A practical problem, considered as an example of ANN application, is the potential for improving the efficiency of identification of large enterprises polluting natural water among many water users in an industrial region. This is made with the use of data on the concentrations of some priority water-polluting metals at the hydrochemical gages in the Iset river near Ekaterinburg City. The neural-network analysis is shown to detect relationships between individual water quality characteristics at nearby gages. This allowed the conclusion that there exist close logistic economic relationships between water users, which help revealing water pollutants by the water footprint produced by plants working in the same branch. It is also shown that the use of ANN opens new ways for determining the contribution of industrial waste discharges to the level of water pollution by substances of dual genesis (natural and technogenic). The reliability of the conclusions is confirmed by the possibility to use the data on a given hydrochemical gage to satisfactorily predict water quality at a gage further downstream.

Enhancing the efficiency of polytetrafluoroethylene-modified silica hydrosols coated solar panels by using artificial neural network and response surface methodology

Abstract In this article, an attempt was made to improve the efficiency of coated solar panels by using artificial neural networks (ANNs) and response surface methodology (RSM). Using the spray coating technique, the glass surface of the photovoltaic solar panel was coated with silicon dioxide nanoparticles incorporated with polytetrafluoroethylene-modified silica sols. Multilayer perceptron with feed-forward back-propagation algorithm was used to develop ANN models for improving the efficiency of the coated solar panels. Out of the 200 sets of data collected, 75% were used for training and 25% were used for testing. On evaluating the models using performance indicators, a four-input technological parameter model (silicon dioxide nanoparticle quantity, coating thickness, surface temperature and solar insolation) with eight neurons in a single hidden layer combination was observed to be the best. The prediction accuracy indicator values of the ANN model were 0.9612 for the coefficient of determination, 0.1971 for the mean absolute percentage error, 0.2317 for the relative root mean square error and 0.00741 for the mean bias error. Using a central composite design model, empirical relationships were developed between input and output responses. The significance of the developed model was ascertained by using analysis of variance, up to a 95% confidence level. For optimization, the RSM was used, and a high efficiency of 17.1% was predicted for the coated solar panel with optimized factors; it was validated to a very high level of predictability. Using interaction and perturbation plots, a ranking of the parameters was done.

PAHs remediation from hazardous waste landfill leachate using fenton, photo – fenton and electro - fenton oxidation processes – performance evaluation under optimized conditions using RSM and ANN

Polycyclic aromatic hydrocharbons (PAHs) are a class of highly toxic pollutants that are highly detrimental to the ecosystem. Landfill leechate emanated from municipal solid waste are reported to constitute significant PAHs. In the present investigation, three Fenton proceses, namely conventional Fenton, photo-fenton and electro-fenton methods have been employed to treat landfill leehcate for removing PAHs from a waste dumpig yard. Response surface methodology (RSM) and artificial neural network (ANN) methodologies were adopted to optimize and validate the conditions for optimum oxidative removal of COD and PAHs. The statistical analysis results showed that all independent variables chosen in the study are reported to have significant influence of the removal effects with P-values <0.05. Sensitivity analysis by the developed ANN model showed that the pH had the highest significance of 1.89 in PAH removal when compared to the other parameters. However for COD removal, H2O2 had the highest relative importance of 1.15, followed by Fe2+ and pH. Under optimal treatment conditions, the photo-fenton and electro-fenton processes showed better removal of COD and PAH compared to the Fenton process. The photo-fenton and electro-fenton treatment processes removed 85.32% and 74.64% of COD and 93.25% and 81.65% of PAHs, respectively. Also the investigations revelaed the presence of 16 distinct PAH compunds and the removal percentage of each of these PAHs are also reported. The PAH treatment research studies are generally limited to the assay of removal of PAH and COD levels. In the present investigation, in addition to the treatment of landfill leachate, particle size distribution analysis and elemental characterization of the resultant iron sludge by FESEM and EDX are reported. It was revealed that elemental oxygen is present in highest percentage, followed by iron, sulphur, sodium, chlorine, carbon and potassium. However, iron percentage can be reduced by treating the Fenton-treated sample with NaOH.

Comments on Das et al., “Optimization of enzymatic saccharification of water hyacinth biomass for bio-ethanol: Comparison between artificial neural network and response surface methodology” [Sustainable Materials and Technologies 3 (2015) 17–28, DOI: 10.1016/j.susmat.2015.01.001

Comments on Das et al., “Optimization of enzymatic saccharification of water hyacinth biomass for bio-ethanol: Comparison between artificial neural network and response surface methodology” [Sustainable Materials and Technologies 3 (2015) 17–28, DOI: 10.1016/j.susmat.2015.01.001

Evaluation of least square support vector machine, generalized regression neural network and response surface methodology in modeling the removal of Levofloxacin and Ciprofloxacin from aqueous solutions using ionic liquid @Graphene oxide@ ionic liquid NC

This study focused on the preparation, characterization of a supported dicationic ionic liquid nanocomposite (IL@GO@IL NC) and employing of it in the adsorption of Levofloxacin (LVX) and Ciprofloxacin (CPF) antibiotic. Also, the nonlinear least square support vector machine (N-LSSVM), generalized regression neural network (GRNN), along with response surface methodology used for modeling the removal of both antibiotics (i.e. LVX and CPF) from aqueous solution. In this work, the experiments designed based on critical parameters such as adsorbent dose, antibiotic concentration, sonication time and temperature. The nonlinear adsorption isotherms and kinetics models as well as thermodynamic used for fitting the experimental data. The statistical results demonstrated that the LSSVM and GRNN model efficiently predicted the antibiotics removal percentage with very high accuracy, respectively. Furthermore, for an optimal adsorption of LVX and CPF by IL@GO@IL NC, the adsorbent dose, antibiotic concentration, temperature and sonication time should be set to 0.23 and 0.35 g, 95.20 and 100.00 mg L-1, 2.10 and 4.60 min, and 45.00 ℃, for LVX and CPF, respectively. Under these optimal conditions, the removal percentages and adsorption capacity of LVX and CPF will be 98.20 and 97.50%, as well as 730 and 630 mg g−1, respectively. Besides, the IL@GO@IL NC efficiency under optimal conditions drops up by 15% after seven cycles for both antibiotics. Therefore, the IL@GO@IL NC may be considered as a promising option in removing fluoroquinolones antibiotics from water.

Neural Network Analysis as a Base of the Future System of Water–Environmental Regulation

The article considers neural-network methods and technologies, which are relatively new even for many researchers and experts, as applied to water–environmental regulation. The efficiency of neural networks in this line of studies is due to their self-training, and the ensuing ability to reveal complex nonlinear relationships between the characteristics under control by data processing instruments, consisting of interrelated neurons. The methodology of artificial neural networks and the features of their functioning are described. Training and methodological examples are given to illustrate their potential use. A practical problem, considered as an example of ANN application, is the potential for improving the efficiency of identification of large enterprises polluting natural water among many water users in an industrial region. This is made with the use of data on the concentrations of some priority water-polluting metals at the hydrochemical gages in the Iset river near Ekaterinburg City. The neural-network analysis is shown to detect relationships between individual water quality characteristics at nearby gages. This allowed the conclusion that there exist close logistic economic relationships between water users, which help revealing water pollutants by the water footprint produced by plants working in the same branch. It is also shown that the use of ANN opens new ways for determining the contribution of industrial waste discharges to the level of water pollution by substances of dual genesis (natural and technogenic). The reliability of the conclusions is confirmed by the possibility to use the data on a given hydrochemical gage to satisfactorily predict water quality at a gage further downstream.

A cascade neural network methodology for fault detection and diagnosis in solar thermal plants

Detecting and isolating faults in collector fields of solar thermal power plants is a crucial and challenging task. The system variables in the collector area are highly coupled, which can lead to a high misclassification rate. For this reason, it becomes necessary to combine knowledge of systems engineering with machine learning techniques that unravel the complex dynamics that govern the systems using historical data. Furthermore, the performance of a solar thermal plant is highly dependent on solar irradiance which changes during the day and is subject to perturbations caused by clouds and other atmospheric conditions. Detecting the fault requires using techniques that cope with the disturbances in solar irradiance.In this work, real irradiance profiles with many types of clouds are used. First, a model-based fault detector is applied, obtaining an accuracy of over 89% for all test irradiances. Then, different machine learning techniques are compared: static neural networks with and without decoupling strategy, dynamic neural networks, dynamic neural networks in cascade, classification trees, random forests, radial basis function networks, and self-organizing maps. The combination of neural networks was the only method that obtained a total accuracy of over 73% and F1-scores over 50% for all the test irradiance profiles.

Application of a parallel physics-informed neural network to solve the multi-body dynamic equations for full-scale train collisions

The prohibitive cost of acquiring data from full-scale train collision experiments limits the applicability of data-driven machine learning methods in train collision simulation. Physics-informed neural networks (PINN) attempt to address this challenge by incorporating physics equations as part of the loss function construction. However, the PINN approach is relatively time-consuming when it comes to solving a large number of physical equations. In this paper, a parallel physics-informed neural network (PPINN) methodology is developed for the solution of multibody dynamics equations to further reduce the computational cost. As well, a PPINN-based framework for engineering applications is proposed, investigating the dynamic responses, absorbed energy, collision forces, and wheel vertical rises of the full-scale train collision. Automatic differentiation and parallelization algorithms are applied to the multi-body dynamic equations including mass, damping, and stiffness matrices, as well as the. The residuals and the initial conditions are included in the loss function. The dynamic responses, absorbed energy, collision forces, and wheel vertical rise of the full-scale train collision are simulated and studied in detail. The results obtained from the PPINN method are in excellent agreement with those obtained from the finite element method (FEM), Newmark-β, and fourth-order Runge–Kutta methods for all four train collision scenarios. Additionally, the PPINN methodology keeps better stability even under large time steps which implies a big potential for computational cost reduction.

Synergistic effects of ternary mixture formulation and process parameters optimization in a sequential approach for enhanced L-asparaginase production using agro-industrial wastes.

A novel ternary mixture of inexpensive and nutrient-rich agro-substrates comprising groundnut de-oiled cake, corn gluten meal, and soybean meal has been explored to enhance the L-asparaginase production in solid-state fermentation. To achieve the aim, a hybrid strategy was implemented by utilizing a combination of a mixture design and artificial neural networks. The study initiated with the judicious selection of the agro-substrates based on their low C/N content in comparison to the control using the CHNS elemental analysis. The mixture composition of soybean meal (49.0%), groundnut de-oiled cake (31.5%), and corn gluten meal (19.5%) were found optimum using the simplex lattice mixture design. The agro-industrial substrates mix revealed synergistic effects on the L-asparaginase production than either of the substrates alone. The maximum L-asparaginase activity of 141.45 ± 5.24IU/gds was observed under the physical process conditions of 70% moisture content, autoclaving period of 30min and 6.0 pH by adopting the machine learning-derived artificial neural network (ANN) methodology. The ANN modeling showed excellent prediction ability with a low mean squared error of 0.7, a low root mean squared error of 0.84, and a high value of 0.99 for regression coefficient. Moisture content (%) was assessed to be the most sensitive process parameter in the global sensitivity analysis. The net outcome from the two sequential optimization designs is the selection of the ideal mixture composition followed by the optimum physical process parameters. The application of the enzyme demonstrated significant cytotoxicity against leukemia cell line and therefore exhibited an anti-cancer effect. The present study reports a novel mixture combination and methodology that can be used to lower the cost and enhance the production of L-asparaginase using an agro-industrial substrate mixture.

Experimental study, prediction modeling, sensitivity analysis, and optimization of rheological behavior and dynamic viscosity of 5W30 engine oil based SiO2/MWCNT hybrid nanofluid

In this paper, the rheological performance and dynamic viscosity of hybrid nanofluid containing SiO2 and multi-walled carbon nanotubes (MWCNTs) nanoparticles (90:10) with 5W30 engine oil as base fluid is experimentally evaluated under different shear rates (SRs) in the range of 50–1000 rpm. The hybrid nanofluid volume fractions (VFs) and temperatures are considered in the ranges of 0.05–1.00 vol% and 5–65 °C, respectively. It was found that the hybrid nanofluid under study behaves as a non-Newtonian fluid. In addition, the calculated power law index was lower than unity, resulting in pseudoplastic features of hybrid nanofluid in all VFs and temperatures. It was observed that the rise of nanofluid temperature from 5 to 65 °C leads to the dynamic viscosity reduction (a 93% decrease in viscosity was observed in a VF of 0.2%), while the increase of nanofluid VF brings about the dynamic viscosity elevation (By increasing VF from 0.05% to 1% at SR of 800 rpm and temperature of 25 °C, the viscosity increases by 29.21%). Based on measured data, an innovative three-variable correlation was established that can more accurately estimate the experimental data than published correlations in the literature. Moreover, the capabilities of GMDH-type neural network (NN) and response surface methodology (RSM) to predict the relative viscosity of the hybrid nanofluid were evaluated. It was concluded that both NN and RSM approaches have a superior ability to forecast the dynamic viscosity behavior of the corresponding hybrid nanofluid, having R2 values of 0.999656 and 0.9955. Furthermore, the optimization was performed and the best solution for achieving the minimum dynamic viscosity with the maximum desirability (1.00) was obtained. Eventually, the dynamic viscosity sensitivity to changes in VF, temperature, and SR was evaluated. It was observed that the dynamic viscosity sensitivity increases as the nanofluid temperature and concentration increase considering a constant SR of 800 rpm.

Optimization of Ultrasonic-Assisted Extraction of α-Glucosidase Inhibitors from Dryopteris crassirhizoma Using Artificial Neural Network and Response Surface Methodology.

Dryopteris crassirhizoma Nakai is a plant with significant medicinal properties, such as anticancer, antioxidant, and anti-inflammatory activities, making it an attractive research target. Our study describes the isolation of major metabolites from D. crassirhizoma, and their inhibitory activities on α-glucosidase were evaluated for the first time. The results revealed that nortrisflavaspidic acid ABB (2) is the most potent α-glucosidase inhibitor, with an IC50 of 34.0 ± 0.14 μM. In addition, artificial neural network (ANN) and response surface methodology (RSM) were used in this study to optimize the extraction conditions and evaluate the independent and interactive effects of ultrasonic-assisted extraction parameters. The optimal extraction conditions are extraction time of 103.03 min, sonication power of 342.69 W, and solvent-to-material ratio of 94.00 mL/g. The agreement between the predicted models of ANN and RSM and the experimental values was notably high, with a percentage of 97.51% and 97.15%, respectively, indicating that both models have the potential to be utilized for optimizing the industrial extraction process of active metabolites from D. crassirhizoma. Our results could provide relevant information for producing high-quality extracts from D. crassirhizoma for functional foods, nutraceuticals, and pharmaceutical industries.

Linear Algebraic Systems Neural Network Solution. Part 1

In recent years, neural networks have become increasingly popular due to their versatility in solving complex problems. One area of interest is their application in solving linear algebraic systems, especially those that are ill-conditioned. The solutions of such systems are highly sensitive to small changes in their coefficients, leading to unstable solutions. Therefore, solving these types of systems can be challenging and require specialized techniques. This article explores the use of neural network methodologies for solving linear algebraic systems, focusing on ill-conditioned systems. The primary goal is to develop a model capable of directly solving linear equations and to evaluate its performance on a range of linear equation sets, including ill-conditioned systems. To tackle this problem, neural network implementing iterative algorithm was built. Error function of linear algebraic system is minimized using stochastic gradient descent. This model doesn’t require extensive training other than tweaking learning rate for particularly large systems. The analysis shows that the suggested model can handle well-conditioned systems of varying sizes, although for systems with large coefficients some normalization is required. Improvements are necessary for effectively solving ill-conditioned systems, since researched algorithm is shown to be not numerically stable. This research contributes to the understanding and application of neural network techniques for solving linear algebraic systems. It provides a foundation for future advances in this field and opens up new possibilities for solving complex problems. With further research and development, neural network models can become a powerful tool for solving ill-conditioned linear systems and other related problems.