Artificial Neuron Network Model Research Articles

Machine learning models for the estimation of monthly mean daily reference evapotranspiration based on cross-station and synthetic data

AbstractThe estimation of reference evapotranspiration (ET0) is important in hydrology research, irrigation scheduling design and water resources management. This study explored the capability of eight machine learning models, i.e., Artificial Neuron Network (ANN), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Extreme Gradient Boosting (XGBoost), Multivariate Adaptive Regression Spline (MARS), Support Vector Machine (SVM), Extreme Learning Machine and a novel Kernel-based Nonlinear Extension of Arps Decline (KNEA) Model, for modeling monthly mean daily ET0 using only temperature data from local or cross stations. These machine learning models were also compared with the temperature-based Hargreaves–Samani equation. The results indicated that the estimation accuracy of these machine learning models differed in various scenarios. The tree-based models (RF, GBDT and XGBoost) exhibited higher estimation accuracy than the other models in the local application. When the station has only temperature data, the MARS and SVM models were slightly superior to the other models, while the ANN and HS models performed worse than the others. When there was no temperature data at the target station and the data from adjacent stations were used instead, MARS, SVM and KNEA were the suitable models. The results can provide a solution for ET0 estimation in the absence of complete meteorological data.

Does selected immunological panel possess the value of predicting the prognosis of early-stage resectable non-small cell lung cancer?

The immune status in the tumor micro-environment of lung cancer is highly informative, and expression status of some biomarkers in the tumor micro-environment may have deep connected statistical characteristics which could indicate immune status and predict related clinical outcomes. In this study, a panel of 9 immune biomarkers are selected for their possible predictive value of clinical outcomes in early-stage resectable non-small cell lung cancer (NSCLC). We introduced a new way of interpreting the expression of Immunohistochemistry-based biomarkers in predicting the prognosis of patients. Immunohistochemistry were performed on surgical samples from 139 patients with NSCLC. Cox regression and subgroup analysis were conducted for the screening of biomarkers that were significant in survival and prognosis. Afterwards, Principle Component Analysis was conducted on screened biomarkers in order to extract characteristic root, which were utilized by an Artificial Neuron Network (ANN) for the prediction of clinical outcomes. Six out of 9 biomarkers were considered significant and were selected for the data analysis. The KMO-Bartlett's Sphericity test was valid (0.658>0.5, P=0.0001). The principle component regression results indicated that the survival was associated with the principle component Z4 (Y=0.316×Z4+2.298, R=0.189, P=0.026). Also, regression-free survival was associated with the principle component Z4 and Z2 (Y=0.314×Z4+0.255×Z2+2.061, R=0.249, P=0.013). Principle components ranking (PCrank) was calculated and after the determination of a cutoff of 0.2, the intergroup comparison of subgrouphigh (PCrank value above 0.2) and subgrouplow (PCrank value below 0.2) was significant in overall survival (OS) (P=0.025). A database search validated our results. Afterwards, the ANN model successfully predicted the clinical outcome, with accuracies of 94.1% and 96.2% respectively in models 1 and 2. The selected immunological panels have promising potential for predicting the prognosis of resectable NSCLC.

An empirical approach of overbreak resistance factor for tunnel blasting

The assessment of overbreak is proposed by means of a novel empirical approach; the ‘overbreak resistance factor’ (ORF), to predict and manage the overbreak phenomenon in tunnel drill-and-blast operations. The proposed ORF is formulated by analysing the relationship between uncontrollable parameters of the overbreak phenomenon, i.e., geological parameters, and the corresponding overbreak measurements. Ninety data sets were collected from the Shin-Hakoishi Tunnel operation in Japan. Initially, an identical weight was applied to all geological parameters to generate ORF subfactors. The contribution of these subfactors to the measured overbreak was analysed through the use of five overbreak prediction artificial neuron network (ANN) models. A sensitivity analysis was conducted on the ANN models to reveal the contributions of input factors to measured overbreak. The discontinuities factors demonstrated the highest influence on overbreak with an overall sensitivity of 55.20%, whereas the strength factors, the weathering factors and the face condition factors showed less sensitivity, at 27.18%, 9.43%, and 8.18% respectively. The sensitivity analysis results were applied back to the initial unweighted data sets to generate a weighted record of subfactors. The ORF values showed a clear inverse proportional relation to the measured overbreak values, through linear regression analysis. Consequently, a five-step ORF prediction chart was developed, which can be directly applied to estimate overbreak in any drill-and-blast tunnel project.

Hot deformation behavior and microstructural evolution of as-quenched 7055 Al alloy fabricated by powder hot extrusion

Isothermal compression tests of powder hot-extruded 7055 Al alloy were carried out on a Gleeble-3180 thermo-simulator at temperatures of 330–450 °C and strain rates of 0.001–1 s−1. A processing map was established by superimposing the power dissipation map over the instability map at a true strain of 0.7. The results show that the instability domain tends to occur at high strain rates and low temperatures, and an optimum processing domain (360–420 °C, 0.001–0.013 s−1) is recommended. The softening mechanism of instability domain is manifested as dislocation accumulation and dynamic recovery (DRV), while that of the recommended domain is mainly coexistence of DRV and partial dynamic recrystallization (DRX). As the temperature increases and the strain rate decreases further, the dominant softening mechanism changes from dynamic recovery (DRV) to continuous dynamic recrystallization (CDRX). Strain-compensated Arrhenius model and artificial neuron network (ANN) model are constructed to predict the flow behavior of the alloy. Compared with the Strain-compensated Arrhenius model, the ANN model shows a better predictability with a lower average absolute relative error of 2.24%, a lower root mean square error of 1.21 MPa, and a higher correlation coefficient of 0.9987.

ANN Approach to Predict the Flow Stress of CMn (Nb-Ti-V) Micro Alloyed Steel

The hot behaviour of micro-alloy steel CMn (Nb-Ti-V) was studied using hot compression tests in a wide range of temperatures (700 to 1050 °C, step 50 °C), deformation rates (0.000794, 0.0029 and 0.01436 s-1) and true deformation rates of 0 to 0.8. Based on experimental stress-deformation data, artificial neuron network (ANN) methods were used to predict flow stress CMn (Nb-Ti-V). The optimal ANN model was developed using Levenberg-Marquardt algorithm, and vas formed with two hidden layers with ten neurons in the first and ten neurons in the second. This model has been shown to be more effective in predicting flow stress and results can also be used in the mathematical simulation of hot metal formation processes.

A nondestructive intelligent approach to real‐time evaluation of chicken meat freshness based on computer vision technique

AbstractIn this study, the capability of a procedure based on combination of computer vision (CV) and artificial intelligence techniques examined for intelligent and nondestructive prediction of chicken meat freshness during the spoilage process at 4°C. The proposed system comprises the following stages: capture images, image preprocessing, image processing, computing channels, feature extraction, feature selection by a hybrid of genetic algorithm (GA) and artificial neuronal network (ANN), and prediction by using ANN. The number of neurons in input layer was determined 33 (selected features) and freshness used as the output. The ideal ANN model was obtained with 33‐10‐1 topology. The high performance of the model was provided with a correlation coefficient of 0.98734 and MSE of 0.002045. The encouraging results of the current study obviously indicated the high potential of CV‐based system combined with an intelligence method as a smart, nondestructive, and reliable technique for online evaluation of chicken meat freshness.Practical applicationDiagnosis and estimation of chicken meat freshness are considered a significant concern in meat quality for consumers. Computer Vision as a novel nondestructive technique can be utilized to evaluate the quality of products. We present the potential of computer vision‐based method as a smart, nondestructive, and reliable method for online prediction of the freshness of chicken meat.

Predictions of equilibrium solubility and mass transfer coefficient for CO2 absorption into aqueous solutions of 4-diethylamino-2-butanol using artificial neural networks

In the present work, artificial neuron network (ANN) based models for predicting equilibrium solubility and mass transfer coefficient of CO 2 absorption into aqueous solutions of high performance alternative 4-diethylamino-2-butanol (DEAB) solvent were successfully developed. The ANN models show an outstanding predictive performance over the predictive correlations proposed in the literature. In order to predict the equilibrium solubility, the ANN model were developed based on three input parameters of operating temperature, concentration of DEAB and partial pressure of CO 2 . An outstanding prediction performance of 2.4% average absolute deviation (AAD) can be obtained (comparing with 7.1–8.3% AAD from the literature). Additionally, a significant improvement on predicting mass transfer coefficient can also be achieved through the developed ANN model with 3.1% AAD (comparing with 14.5% AAD from the existing semi-empirical model). The mass transfer coefficient is considered to be a function of liquid flow rate, liquid inlet temperature, concentration of DEAB, inlet CO 2 loading, outlet CO 2 loading, concentration of CO 2 along the height of the column.

New silica–titania based polymeric hybrid materials for the removal of Cu(II) ions from their aqueous solutions

Large surface area, high porosity and good mechanical strength are some of the attributes of the polymer–inorganic hybrid materials those make these attractive candidates for use as adsorbents. New silica/titania–based polymer–inorganic hybrid material was synthesized via the sol–gel process. While SiO2 and TiO2 were used as the inorganic components, poly(2–Hydroxypropyl methacrylate) was used as the polymeric component. The material was well characterized by FTIR, SEM, TEM and BET. The synthesized hybrid material was used as adsorbent for the removal of Cu(II) ions from their aqueous solutions. Parametric study of the factors affecting adsorption was carried out to find the optimum adsorption conditions. Different kinetic and isotherm models applied show that the adsorption best follows pseudo–second order kinetics and Langmuir isotherm. The maximum adsorption capacity of the HPMA/SiO2/TiO2 hybrid material was found to be 192.23mgg.−1 In order to see the validity of the experimental procedure and results obtained, artificial neuron network model was employed to predict the adsorption capacity values following the experimental adsorption parametric frame. There is a close agreement with the experimental and the predicted values as very high values of R2 of 0.99957 and 0.99961 were obtained for the predicted vs. experimental values with training and testing data set.

Integrated ANN-LWPA for cutting parameter optimization in WEDM

In this paper, we present a new approach to determinate cutting parameters in wire electrical discharge machining (WEDM), integrated artificial neuron network (ANN), and wolf pack algorithm based on the strategy of the leader (LWPA). The cutting parameters considered in this paper are pulse-on, current, water pressure, and cutting feed rate. Models of the effects of the four parameters on machining time (Tp), machining cost (Cp), and surface roughness (Ra) are mathematically constructed. An ANN-LWPA integration system with multiple fitness functions is proposed to solve the modelling problem. By using the proposed approach, this study demonstrates that Tp, Cp, and Ra can be estimated at 164.1852 min, 239.5442 RMB, and 1.0223 μm in single objective optimization, respectively. For example, as for Ra, integrated ANN-LWPA has optimized the Ra value by the reduction of 0.1337 μm (11.6 %), 0.3377 μm (24.8 %), and 0.105 μm (10.3 %) compared to experimental data, regression model, and ANN model, respectively. Consequently, the ANN-LWPA integration system boasts some advantages over decreasing the value of fitness functions by comparison with the experimental regression model, ANN model, and conventional LWPA result. Moreover, the proposed integration system can be also utilized to obtain multiple solutions by uniform design-based exploration. Therefore, in order to solve complex machining optimization problems, an intelligent process scheme could be integrated into the numeric control system of WEDM.

Mosquito vector monitoring system based on optical wingbeat classification

We developed an automatic mosquito classification system which consists of an infrared recording device for profiling the wingbeat of the in-flight mosquito species and a machine learning model for classifying the gender, genus, and species of the incoming mosquitoes by the signatures of their wingbeats.The recording device is a set of infrared emitters and receivers, which are attached to the wall of an apparatus. When the winged subject enters the apparatus, its flapping wings block the infrared beam from the emitters intermittently such that the receivers convert the wingbeat to the electrical waveform. To classify the incoming subjects, we proposed a machine learning method, which is the Gaussian mixture model trained using the expectation-maximization algorithm (EM-GMM), and compared it with the previously proposed algorithms, including the artificial neuron network model (ANN) and the nearest neighbor model.To assess the performance of the system, we used the living male and female Aedes albopictus, Aedes aegypti and Culex quinquefasciatus. The results show that the accuracies of the proposed system are above 80% on identifying the gender and genus of the mosquitoes, with the precisions above 80% and 70%, respectively. The results also suggest that the EM-GMM algorithm outperforms the other two algorithms on the accuracy and precision of the classification of the classes of mosquitoes. In addition to the evaluation of the performance of the system, we also found that certain classes of mosquitoes share similar wingbeat characteristics, which implies that the distinctive wingbeat characteristics should be considered for the optimal accuracy of the classification of the insects of interest.

Estimation of surface upward longwave radiation from MODIS and VIIRS clear-sky data in the Tibetan Plateau

Surface upward longwave radiation (SULR) is one critical component of the surface energy balance and is closely related to the surface temperature fields. The SULR with the moderate spatial resolution and appropriate precision in the Tibetan Plateau, a large and complex cryosphere, can enhance our understanding of the land surface processes, atmospheric circulations and even global climate change. The purpose of this paper is to estimate SULR from top-of-atmosphere (TOA) radiances of Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS). A hybrid method is proposed that combines extensive radiative transfer simulation and the Artificial Neuron Network (ANN) statistical model. For the sake of ensuring the availability of the simulation database, the eligible profiles from MODIS atmospheric profiles product were screened according to five criteria. And then the analyses on the important parameters of the simulation dataset were performed to check the integrality of this dataset. Moreover, the variables' importance was analyzed, and TOA radiances of MODIS channels 29, 31, 32 and 33 (or VIIRS channels M14, M15 and M16) and sensor view zenith angle (VZA) were selected to retrieve SULR. The ANN-based retrieval model was created finally based on this simulation dataset. The evaluation showed good agreements for both MODIS and VIIRS training and testing datasets with R2 greater than 0.98. The MODIS SULR was validated using two-year ground-measured data at seven sites in the Tibetan Plateau with R2 of 0.886, root-mean-square error (RMSE) of 26.985Wm−2 and mean bias error (MBE) of 10.812Wm−2 as a whole. And the relevance was obvious from the overall trend and distribution though the discrepancy was relatively large when directly comparing SULR from VIIRS and MODIS data. The validation results showed that the ANN model is a good nonlinear model to retrieve SULR with the TOA radiances and VZA.

Patterns Prediction of Chemotherapy Sensitivity in Cancer Cell lines Using FTIR Spectrum, Neural Network and Principal Components Analysis.

Drug resistance enables cancer cells to break away from cytotoxic effect of anticancer drugs. Identification of resistant phenotype is very important because it can lead to effective treatment plan. There is an interest in developing classifying models of resistance phenotype based on the multivariate data. We have investigated a vibrational spectroscopic approach in order to characterize a sensitive human ovarian cell line, A2780, and its cisplatin-resistant derivative, A2780-cp. In this study FTIR method have been evaluated via the use of principal components analysis (PCA), ANN (artificial neuronal network) and LDA (linear discriminate analysis). FTIR spectroscopy on these cells in the range of 400-4000 cm(-1) showed alteration in the secondary structure of proteins and a CH stretching vibration. We have found that the ANN models correctly classified more than 95% of the cell lines, while the LDA models with the same data sets could classify 85% of cases. In the process of different ranges of spectra, the best classification of data set in the range of 1000-2000 cm(-1) was done using ANN model, while the data set between 2500-3000 cm(-1) was more correctly classified with the LDA model. PCA of the spectral data also provide a good separation for representing the variety of cell line spectra. Our work supports the promise of ANN analysis of FTIR spectrum as a supervised powerful approach and PCA as unsupervised modeling for the development of automated methods to determine the resistant phenotype of cancer classification.

Network approaches for expert decisions in sports

This paper focuses on a model comparison to explain choices based on gaze behavior via simulation procedures. We tested two classes of models, a parallel constraint satisfaction (PCS) artificial neuronal network model and an accumulator model in a handball decision-making task from a lab experiment. Both models predict action in an option-generation task in which options can be chosen from the perspective of a playmaker in handball (i.e., passing to another player or shooting at the goal). Model simulations are based on a dataset of generated options together with gaze behavior measurements from 74 expert handball players for 22 pieces of video footage. We implemented both classes of models as deterministic vs. probabilistic models including and excluding fitted parameters. Results indicated that both classes of models can fit and predict participants’ initially generated options based on gaze behavior data, and that overall, the classes of models performed about equally well. Early fixations were thereby particularly predictive for choices. We conclude that the analyses of complex environments via network approaches can be successfully applied to the field of experts’ decision making in sports and provide perspectives for further theoretical developments.

NL MIND-BEST: A web server for ligands and proteins discovery—Theoretic-experimental study of proteins of Giardia lamblia and new compounds active against Plasmodium falciparum

There are many protein ligands and/or drugs described with very different affinity to a large number of target proteins or receptors. In this work, we selected Ligands or Drug-target pairs (DTPs/nDTPs) of drugs with high affinity/non-affinity for different targets. Quantitative Structure–Activity Relationships (QSAR) models become a very useful tool in this context to substantially reduce time and resources consuming experiments. Unfortunately most QSAR models predict activity against only one protein target and/or have not been implemented in the form of public web server freely accessible online to the scientific community. To solve this problem, we developed here a multi-target QSAR (mt-QSAR) classifier using the MARCH-INSIDE technique to calculate structural parameters of drug and target plus one Artificial Neuronal Network (ANN) to seek the model. The best ANN model found is a Multi-Layer Perceptron (MLP) with profile MLP 20:20-15-1:1. This MLP classifies correctly 611 out of 678 DTPs (sensitivity=90.12%) and 3083 out of 3408 nDTPs (specificity=90.46%), corresponding to training accuracy=90.41%. The validation of the model was carried out by means of external predicting series. The model classifies correctly 310 out of 338 DTPs (sensitivity=91.72%) and 1527 out of 1674 nDTP (specificity=91.22%) in validation series, corresponding to total accuracy=91.30% for validation series (predictability). This model favorably compares with other ANN models developed in this work and Machine Learning classifiers published before to address the same problem in different aspects. We implemented the present model at web portal Bio-AIMS in the form of an online server called: Non- Linear MARCH-INSIDE Nested Drug- Bank Exploration & Screening Tool ( NL MIND-BEST), which is located at URL: http://miaja.tic.udc.es/Bio-AIMS/NL-MIND-BEST.php. This online tool is based on PHP/HTML/Python and MARCH-INSIDE routines. Finally we illustrated two practical uses of this server with two different experiments. In experiment 1, we report by first time Quantum QSAR study, synthesis, characterization, and experimental assay of antiplasmodial and cytotoxic activities of oxoisoaporphine alkaloids derivatives as well as NL MIND-BEST prediction of potential target proteins. In experiment 2, we report sampling, parasite culture, sample preparation, 2-DE, MALDI-TOF, and -TOF/TOF MS, MASCOT search, MM/MD 3D structure modeling, and NL MIND-BEST prediction for different peptides a new protein of the found in the proteome of the human parasite Giardia lamblia, which is promising for anti-parasite drug-targets discovery.

Using entropy of drug and protein graphs to predict FDA drug-target network: Theoretic-experimental study of MAO inhibitors and hemoglobin peptides from Fasciola hepatica

There are many drugs described with very different affinity to a large number of receptors. In this work, we selected Drug-Target pairs (DTPs/nDTPs) of drugs with high affinity/non-affinity for different targets like proteins. Quantitative Structure–Activity Relationships (QSAR) models become a very useful tool in this context to substantially reduce time and resources consuming experiments. Unfortunately, most QSAR models predict activity against only one protein. To solve this problem, we developed here a multi-target QSAR (mt-QSAR) classifier using the MARCH-INSIDE technique to calculate structural parameters of drug and target plus one Artificial Neuronal Network (ANN) to seek the model. The best ANN model found is a Multi-Layer Perceptron (MLP) with profile MLP 32:32–15–1:1. This MLP classifies correctly 623 out of 678 DTPs (Sensitivity = 91.89%) and 2995 out of 3234 nDTPs (Specificity = 92.61%), corresponding to training Accuracy = 92.48%. The validation of the model was carried out by means of external predicting series. The model classifies correctly 313 out of 338 DTPs (Sensitivity = 92.60%) and 1411 out of 1534 nDTP (Specificity = 91.98%) in validation series, corresponding to total Accuracy = 92.09% for validation series (Predictability). This model favorably compares with other LDA and ANN models developed in this work and Machine Learning classifiers published before to address the same problem in different aspects. These mt-QSARs offer also a good opportunity to construct drug–protein Complex Networks (CNs) that can be used to explore large and complex drug–protein receptors databases. Finally, we illustrated two practical uses of this model with two different experiments. In experiment 1, we report prediction, synthesis, characterization, and MAO-A and MAO-B pharmacological assay of 10 rasagiline derivatives promising for anti-Parkinson drug design. In experiment 2, we report sampling, parasite culture, SEC and 1DE sample preparation, MALDI-TOF MS and MS/MS analysis, MASCOT search, MM/MD 3D structure modeling, and QSAR prediction for different peptides of hemoglobin found in the proteome of the human parasite Fasciola hepatica; which is promising for anti-parasite drug targets discovery.

Comparison between back propagation neural network and regression models for the estimation of pigment content in rice leaves and panicles using hyperspectral data

Two nitrogen experiments on rice were conducted in 2002, and the reflectances (350 to 2500 nm) and pigment contents (chlorophylls a and b, total chlorophylls and carotenoids) for leaf and panicle samples at different growth stages were measured in the laboratory. After performing an outlier analysis, the number of samples were 843 for leaves and 188 for panicles. Absorption features at 430, 460, 470, 640 and 660 nm for different pigments, and the relative reflectance of the green peak around 550 nm calculated by the continuum‐removed method, as well as the red edge position (REP) of rice leaves and panicles were selected as the independent variables, and measured pigment contents were selected as the dependent variables. Then, back propagation neural network (BPN) models, a kind of artificial neuron network (ANN), and multivariate linear regression models (MLR) were trained and tested. The main objective of this study was to compare the predictive ability of the ANN models to that of the MLR models in estimating the content of pigments in rice leaves and panicles. Results showed that all BPN models gave higher coefficients of determination (R2) and lower absolute errors (ABSEs) and root mean squared errors (RMSEs) than the corresponding MLR models, in both calibration and validation tests. Further significance tests by paired t tests and bootstrapping algorithms indicated that most of the BPN models outperformed the MLR models. When trained by combination data that did not meet the assumption of normal distribution, the BPN models appeared to not only have a better learning ability, but also had a more accurate predictive power than the MLR models. The estimation of leaf pigments was more accurate than that of panicle pigments, independent of which model was used.

Wavelet and artificial neural network analyses of tide forecasting and supplement of tides around Taiwan and South China Sea

In multi-resolution analysis (MRA) by wavelet function Daubechies (db), we decompose the signal in two parts, the low and high-frequency contents. We remove the high-frequency content and reconstruct a new “de-noise” signal by using inverse wavelet transform. The calculation of tidal constituent phase-lags was made to determine the input and output data patterns used in building network structure of Artificial Neuron-Network (ANN) model. The “de-noise” signal was, then, used as the input data to improve the forecasting accuracy of the ANN model. The wavelet spectrum, conventional energy spectrum (fast Fourier transform, FFT), and harmonic analysis were used to analyze the characteristics of tidal data. Using only a very short-period data as a training data set in Artificial Neuron-Network Back-Propagate (ANN-BP) model, the developed ANN+Wavelet model can accurately predict or supply the missing tide data for a long period (1–5 years). The results also show that the concept of tidal constituent phase-lags can improve ANN model of tidal forecasting and data supplement. The addition of the wavelet analysis to ANN method can prominently improve the prediction quality.