Principal Component Analysis-artificial Neural Network Research Articles

Principal component analysis–artificial neural network-based model for predicting the static strength of seasonally frozen soils

Seasonally frozen soils are exposed to freeze‒thaw cycles every year, leading to mechanical property deterioration. To reasonably describe the deterioration of soil under different conditions, machine learning (ML) technology is used to establish a prediction model for soil static strength. Six key influencing factors (moisture content, compaction degree, confining pressure, freezing temperature, number of freeze‒thaw cycles and thawing duration) are included in the modelling database. The accuracy of three typical ML algorithms (support vector machine (SVM), random forest (RF) and artificial neural network (ANN)) is compared. The results show that the ANN outperforms the SVM and RF. Principal component analysis (PCA) is combined with the ANN, and the PCA–ANN algorithm is proposed, which further improves the prediction accuracy. The deterioration of soil static strength is systematically researched using the PCA–ANN algorithm. The results show that the soil static strength decreased considerably after the first several freeze‒thaw cycles before the strength plateau occurred, and the strength reduction increased significantly with increasing moisture content and compaction degree. The PCA–ANN model can generate a reasonable prediction for the static strength or other soil properties of seasonally frozen soil, which will provide a scientific reference for practical engineering.

Hybrid Machine Learning Model for Face Recognition Using SVM

Face recognition systems have enhanced human-computer interactions in the last ten years. However, the literature reveals that current techniques used for identifying or verifying faces are not immune to limitations. Principal Component Analysis-Support Vector Machine (PCA-SVM) and Principal Component Analysis-Artificial Neural Network (PCA-ANN) are among the relatively recent and powerful face analysis techniques. Compared to PCA-ANN, PCA-SVM has demonstrated generalization capabilities in many tasks, including the ability to recognize objects with small or large data samples. Apart from requiring a minimal number of parameters in face detection, PCA-SVM minimizes generalization errors and avoids overfitting problems better than PCA-ANN. PCA-SVM, however, is ineffective and inefficient in detecting human faces in cases in which there is poor lighting, long hair, or items covering the subject's face. This study proposes a novel PCA-SVM-based model to overcome the recognition problem of PCA-ANN and enhance face detection. The experimental results indicate that the proposed model provides a better face recognition outcome than PCA-SVM.

Mark-Spectra: A convolutional neural network for quantitative spectral analysis overcoming spatial relationships

Spectral analysis is one of the most important and widely used methods for chemometrics in the field of agriculture, and convolutional neural network (CNN) models have achieved excellent performance on spectral analysis. The critical drawback of the CNN approach is that it preserves the spatial relationships among adjacent wavelengths, which contribute to collinearity and redundancies rather than relevant effective information. To confirm this observation, the distribution of characteristic wavelengths extracted by different methods (include F-test, importance weights, and CNN) are visualized in this paper. A convolutional neural network for quantitative spectral analysis, named Mark-Spectra, is presented to overcome spatial relationships and to improve the model performance. A layer (Mark layer) is introduced as part of Mark-Spectra, which is used to overcome spatial relationship of raw spectral data. Mark-Spectra model is compared with three CNN models using three open accessed visible and near infrared spectroscopic datasets (corn, wheat and soil). Mark-Spectra model outperforms the other three convolutional neural network models on two datasets (except dataset of wheat, due to lesser number of features), and it cost much less training time than the others. In addition, this paper compares Mark-Spectra with two classical neural network-based algorithms, principal component analysis – artificial neural network (PCA-ANN) and extreme learning machine (ELM). Mark-Spectra performed best in soil dataset, and ELM performed best in corn and wheat datasets, respectively. These results can illustrate that Mark-Spectra is still limited with the characteristic of raw spectral data (e.g., the number of samples and features), which is a fundamental fact of deep learning-based methods, but it performed better than the other CNN models and reduced the dependence of sample size due to overcoming spatial relationships.

A novel PCA-FA-ANN based hybrid model for prediction of fluoride

Fluoride plays an essential role in terms of the health of human beings. Persistent exposure to fluoride, which is present in drinking water mainly, may result in dental, skeletal, and non-skeletal fluorosis. However, drinking water, being consumed presently, is not sufficient to indicate the degree of exposure to fluoride. The existing literature indicates that nails can be used as indicative (biomarkers) of not only to exposure of fluoride but also the degree of the same. However, because of differential metabolism rate depending on a number of factors like age, gender, nutritional status, water characteristics, etc., exposure to fluoride is not easily detectable in human beings by just testing the fluoride content in nails. Moreover, due to sensitive chemical analysis and lack of facilities, it is difficult to identify the exact concentration of fluoride in nails. The objective of this study is to identify the significant parameters that affect the fluoride content in nail samples. Apart from laboratories test, the application of different artificial intelligence (AI) methods are used for the prediction of fluoride in nails, which will help to identify the degree of fluoride exposure to children, females, and males. The field study covers a collection of 2401 nail samples from eight districts of Rajasthan, India. The samples were taken of different age groups and genders. 1024 water samples were also collected from the same households from where the nails samples were collected. These water and nail samples were tested in laboratories at Birla Institute of Technology and Science, Pilani, India. This data set was used to train, test, and validate the results using AI techniques. The proposed hybrid model (HM) combines Principal Component Analysis (PCA), Firefly Algorithm (FA), and Artificial Neural Network (ANN) to predict fluoride concentration in nails. The model performance was assessed by using different errors and coefficient of relationship. A comparative analysis of the proposed HM was done against ANN and PCA–ANN models. Sensitivity analysis was performed to study the importance of input parameters on the output. Results show that the PCA–FA–ANN–AFpE model having four input parameters (age, fluoride, pH, and electrical conductivity), displayed the best performance with high R2 values of 0.97. The study generates a novel methodology for predicting fluoride concentration in nails, which can be used for the prediction of hotspots in terms of continuous exposure to fluoride.

Low‐resolution fiber‐optic Raman spectroscopy for bladder cancer diagnosis: A comparison study of varying laser power, integration time, and classification methods

AbstractIn our previous work, we have demonstrated the great diagnostic potential of a low‐resolution Raman sensing system for bladder cancer through ex vivo experiments. Before forwarding this technique into clinical applications, the system's performance under different experimental conditions must be thoroughly understood. In this paper, a comparison study of this system under different experimental conditions and post‐experiment analysis methods is presented. The different experiment conditions includes two major parts: (a) varying the incident laser power at sample from 30 to 150 mW (30‐mW interval) with fixed integration time of 1 s; (b) varying integration time of 1 s, 2 s, 3 s, and 5 s, with a fixed incident laser power of 150 mW. A total number of 2,916 spectra were collected on 42 bladder tissue specimens under different experimental conditions. Three principal component analysis (PCA)‐based classification methods, including linear discriminant analysis (LDA), support vector machine (SVM), and artificial neural network (ANN), are used in this study for comparison. Results show that increasing the incident laser power has little influence on the overall prediction accuracy; increasing the integration time from 1 to 5 s has a clear improvement on the prediction accuracy; PCA–ANN outperforms PCA–LDA and PCA–SVM consistently under the parameter settings in this study.

Determination of the superficial citral content on microparticles: An application of NIR spectroscopy coupled with chemometric tools

This work evaluates near-infrared (NIR) spectroscopy coupled with chemometric tools for determining the superficial content of citral (SCCt) on microparticles. To perform this evaluation, using spray drying, citral was encapsulated in a matrix of dextrin using twelve combinations of citral:dextrin ratios (CDR) and inlet air temperatures (IAT). From each treatment, six samples were extracted, and their SCCt and NIR absorption spectral profiles were measured. Then, the spectral profiles, pretreated and randomly divided into modeling and validation datasets, were used to build the following prediction models: principal component analysis-multilinear regression (PCA-MLR), principal component analysis-artificial neural network (PCA-ANN), partial least squares regression (PLSR) and an artificial neural network (ANN). During the validation stage, the models showed R2 values from 0.73 to 0.96 and a root mean squared error (RMSE) range of [0.061–0.140]. Moreover, when the models were compared, the full and optimized ANN models showed the best fits. According to this study, NIR coupled with chemometric tools has the potential for application in determining SCCt on microparticles, particularly when using ANN models.

A new wavelet conjunction approach for estimation of relative humidity: wavelet principal component analysis combined with ANN

Relative humidity (RH) has an important effect on precipitation, especially in arid and semiarid regions. Prediction of RH has been the focus of attention of climate change researchers as well. In this investigation, the accuracy of six intelligent models, including an artificial neural network (ANN), a co-active neuro-fuzzy inference system (CANFIS), principal component analysis (PCA) combined with ANN (PCA–ANN) and three hybrid wavelet-artificial intelligence models, including WANN, WCANFIS and WPCA–ANN, was evaluated in daily RH prediction. Thirty weather stations located in different climates in Iran for the period 2000–2010 were selected for the evaluation and comparison of these models. The performance of each model was evaluated using correlation coefficient (r) and normal root mean square error (NRMSE). Based on the statistical evaluation criteria, the accuracy ranks of the six models were: WPCA–ANN, WCANFIS, WANN, PCA–ANN, ANN and CANFIS. The results indicated that the WPCA–ANN model was the optimal model for estimation of RH, and the range of NRMSE and r values were from 0.009 to 0.080 and from 0.996 to 0.999, respectively. Overall, WPCA–ANN is a new approach that can be successfully applied to predict RH with a high accuracy.

Combined Unfolded Principal Component Analysis and Artificial Neural Network for Determination of Ibuprofen in Human Serum by Three-Dimensional Excitation-Emission Matrix Fluorescence Spectroscopy.

This study describes a simple and rapid approach of monitoring ibuprofen (IBP). Unfolded principal component analysis-artificial neural network (UPCA-ANN) andexcitation-emission spectra resulted from spectrofluorimetry method were combined to develop new model in the determination of IBF in human serum samples.Fluorescencelandscapes with excitation wavelengths from 235 to 265 nm and emission wavelengths in the range 300-500 nm were obtained. The figures of merit for the developed model were evaluated. High performance liquid chromatography (HPLC) technique was also used as a standard method. Accuracy of the method was investigated by analysis of the serum samples spiked with various concentration of IBF and an average relative error of prediction of 0.18% was obtained. The results indicated that the proposed method is an interesting alternative to the traditional techniques normally used for determination of IBF such as HPLC.

SnO2 nanoparticle-loaded activated carbon for simultaneous removal of Acid Yellow 41 and Sunset Yellow; derivative spectrophotometric, artificial neural network and optimization approach

The simultaneous adsorption of Acid Yellow 41 (AY41) and Sunset Yellow (SY) onto SnO2 nanoparticle-loaded activated carbon (SnO2–NP–AC with very high BET surface area of 1278.71m2g−1) was investigated. To overcome the severe dyes spectral overlapping, derivative spectrophotometric method and principal component analysis-artificial neural network (PCA-ANN) were successfully applied for the simultaneous determination of AY41 and SY in their binary solutions. By using central composite design (CCD) under response surface methodology, the effects of variables such as contact time, adsorbent dosage, pH, AY41 concentration and SY concentration on responses such as binary dyes removal percentages were examined. Optimal values were found to be 17.9min, 0.024g, 3.1, and 15.9mgL−1 and 18.7mgL−1, respectively. In binary solutions, the best fit to modified–extended Langmuir isotherm was obtained for the whole concentration range. In binary solutions, a synergism was observed for the AY41 and SY dyes adsorption onto SnO2–NP–AC. The adsorption rates at various times were analyzed. It indicated a pseudo-second-order kinetic model for the adsorption of both dyes.

Predicting the Amount of Municipal Solid Waste via Hybrid Principal Component Analysis-Artificial Neural Network Approach

Accurate prediction of the amount of municipal solid waste (MSW) is crucial for designing and programming MSW management systems. Reliable estimation of MSW is difficult since many variables such as socio-economic characteristics, climatic factors and standard of living affect it. A number of studies used artificial neural network (ANN) to predict MSW. However, due to the large number of input variables to the ANN, it could not not perform well and generally encountered overfitting. This study takes advantage of the principal component analysis (PCA) technique to reduce the number of input variables to the ANN model in order to overcome the overfitting problem. The proposed PCA-ANN approach is used to predict the weight of MSW in the province of Mashhad, Iran. The utilized experimental data in this study are obtained from the Recycling Organization of Mashhad Municipality archive (http://www.wmo.mashhad.ir). It is found that the PCA approach can successfully decrease the number of input variables from thirteen to eight. The PCA-ANN model (with eight input variables) outperforms ANN (with thirteen input variables) and provides more accurate estimates of MSW as it mitigates the overfitting problem associated with ANN. The root-mean-square-error (RMSE) of MSW estimates reduces from 499000 Kg to 448000 Kg by using the PCA-ANN model instead of ANN.

Application of factor analysis in a large‐scale industrial wastewater treatment plant simulation using principal component analysis–artificial neural network hybrid approach: (Case Study: Fajr Industrial Wastewater Treatment Plant, Mahshahr, Iran)

The effective operation of industrial wastewater treatment plants is quite complicated due to having diverse qualitative and quantitative variations in their effluent characteristics during a day. In this article, we take full advantages of well‐known prediction models to acquire an applicable and constructive operation over industrial treatment plants. We combine multilayer perceptron feed forward neural networks with Levenberg–Marquardt training function (Trainlm) and principal component analysis method to estimate pH, chemical oxygen demand, total dissolved solid, Cl−, turbidity, and achieve appropriate operation of Fajr petrochemical industrial treatment plant for the first time in Iran. Moreover, factor analysis approach was applied to determine the paramount input parameters of the models to reduce the parameters' dimension. Mean square error, root mean square error, and correlation coefficient (R) were used for evaluating the performance of the models. Results indicate that correlation coefficients (R) in the range of 0.8–0.94 showed excellent accuracy of the models in estimating qualitative profile of wastewater. Simulation of a whole treatment plant, better prediction of parameters, and proposing a new hybrid model could be some advantages of this study. © 2015 American Institute of Chemical Engineers Environ Prog, 34: 1322–1331, 2015

An Enhanced Empirical Method on Choosing the Highest Principal Features and the Number of Hidden Neurons in Principal Component Analysis-Artificial Neural Network Face Recognition based System

ABSTRACT With fast evolving technology, it is necessary to design an efficient security system which can detect unauthorized access on any system. It‟s needed to implement an extremely secure, economic and perfect system for face recognition that can protect systems from unauthorized access. So, in this paper, a robust face recognition system approach is proposed for feature extraction using Principal Component Analysis, and recognition using Feed Forward Back Propagation Neural Network. The proposed approach gave better results in all aspects including recognition rate, training time, elapsed time and mean square error. The paper shows that when using 80% of the dataset for training, the proposed system can achieve up to 97.5% recognition rate if correct number of input principal features to the classifier, learning rate, momentum, and number of hidden neurons are used. This algorithm is implemented using Matlab software, and tolerable error value used is superiorly chosen as MSE= 0.0004.

Development of artificial intelligence based NO2 forecasting models at Taj Mahal, Agra

The statistical regression and specific computational intelligence based models are presented in this paper for the forecasting of hourly NO2 concentrations at a historical monument Taj Mahal, Agra. The model was developed for the purpose of public health oriented air quality forecasting. Last ten–year air pollution data analysis reveals that the concentration of air pollutants increased significantly. It is also observed that the pollution levels are always higher during the months of November at around Taj Mahal, Agra. Therefore, the hourly observed data during November were used in the development of air quality forecasting models for Agra, India. Firstly, multiple linear regression (MLR) was used for building an air quality–forecasting model to forecast the NO2 concentrations at Agra. Further, a novel approach, based on regression models, principal component analysis (PCA) was analyzed to find the correlations of different predictor variables between meteorology and air pollutants. Then, the significant variables were taken as the input parameters to propose the reliable physical artificial neural network (ANN)-multi layer perceptron model for forecasting of air pollution in Agra. MLR and PCA–ANN models were evaluated through statistical analysis. The correlation coefficients (R) were 0.89 and 0.91 respectively, for PCA–ANN and were 0.69 and 0.89 respectively for MLR in the training and validation periods. Similarly, the values of normalized mean square error (NMSE), index of agreement (IOA) and fractional bias (FB) were in good agreement with the observed values. It was concluded that PCA–ANN model performs better and can be used for forecasting air pollution at Taj Mahal, Agra.

Competitive adsorption of methylene blue and brilliant green onto graphite oxide nano particle following: Derivative spectrophotometric and principal component-artificial neural network model methods for their simultaneous determination

In this work, the competitive adsorption of methylene blue (MB) and brilliant green (BG) onto graphite oxide (GO) nanoparticles followed by their accurate and reproducible determination by second order derivative spectrophotometry (SODS) and principal component–artificial neural network model (PCA–ANN) model has been studied. The evaluation of kinetic and isotherm studies was investigated at optimum experimental conditions set as pH=7.0, 8mg of GO and 14min contact time in binary systems. The equilibrium amounts of MB and BG dyes in binary mixture adsorbed onto GO-NP has opposite correlation with their initial concentration. Principal component analysis (PCA) used to minimize the dimensionality of large data sets via reducing the number of spectral data by a three-layered feed-forward artificial neural network (ANN) trained by Levenberg–Marquardt back-propagation algorithm. The ANN model was able to predict the concentrations of both dyes in mixtures with a tangent sigmoid transfer function (tansig) at hidden layer with 20 neurons and a linear transfer function (purelin) at output layer. Several isotherm models were applied to experimental data and the isotherm constants were calculated for BG and MB dyes. Among the applied models, the extended Freundlich isotherm model adequately predicts the multi-component adsorption equilibrium data at moderate ranges of concentration.

Principal component analysis-artificial neural network and genetic algorithm optimization for removal of reactive orange 12 by copper sulfide nanoparticles-activated carbon

In this study a green approach described for the synthesis of copper sulfide nanoparticles loaded on activated carbon (CuS-NP-AC) and usability of it for the removal of reactive orange 12 (RO-12). This material was characterized using instruments such as scanning electron microscopy (SEM) and X-ray diffraction (XRD). The effects of variables were optimized using Principal component analysis-artificial neural network (PCA-ANN). Fitting the experimental equilibrium data shows the suitability of the Langmuir isotherm. The small amount of proposed adsorbent (0.017g) is applicable for successful removal of RO-12 (RE>95%) in short time (31.09min) with high adsorption capacity (96.9mgg−1)

Classification of time-of-flight secondary ion mass spectrometry spectra from complex Cu–Fe sulphides by principal component analysis and artificial neural networks

Artificial neural network (ANN) and a hybrid principal component analysis-artificial neural network (PCA-ANN) classifiers have been successfully implemented for classification of static time-of-flight secondary ion mass spectrometry (ToF-SIMS) mass spectra collected from complex Cu–Fe sulphides (chalcopyrite, bornite, chalcocite and pyrite) at different flotation conditions. ANNs are very good pattern classifiers because of: their ability to learn and generalise patterns that are not linearly separable; their fault and noise tolerance capability; and high parallelism. In the first approach, fragments from the whole ToF-SIMS spectrum were used as input to the ANN, the model yielded high overall correct classification rates of 100% for feed samples, 88% for conditioned feed samples and 91% for Eh modified samples. In the second approach, the hybrid pattern classifier PCA-ANN was integrated. PCA is a very effective multivariate data analysis tool applied to enhance species features and reduce data dimensionality. Principal component (PC) scores which accounted for 95% of the raw spectral data variance, were used as input to the ANN, the model yielded high overall correct classification rates of 88% for conditioned feed samples and 95% for Eh modified samples.

Application of silver nanoparticles and principal component-artificial neural network models for simultaneous determination of levodopa and benserazide hydrochloride by a kinetic spectrophotometric method

A multicomponent analysis method based on principal component analysis-artificial neural network model (PC-ANN) is proposed for the simultaneous determination of levodopa (LD) and benserazide hydrochloride (BH). The method is based on the reaction of levodopa and benserazide hydrochloride with silver nitrate as an oxidizing agent in the presence of PVP and formation of silver nanoparticles. The reaction monitored at analytical wavelength 440 nm related to surface plasmon resonance band of silver nanoparticles. Differences in the kinetic behavior of the levodopa and benserazide hydrochloride were exploited by using principal component analysis, an artificial neural network (PC-ANN) to resolve concentration of analytes in their mixture. After reducing the number of kinetic data using principal component analysis, an artificial neural network consisting of three layers of nodes was trained by applying a back-propagation learning rule. The optimized ANN allows the simultaneous determination of analytes in mixtures with relative standard errors of prediction in the region of 4.5 and 6.3 for levodopa and benserazide hydrochloride respectively. The results show that this method is an efficient method for prediction of these analytes.

New spectral approaches to the simultaneous quantitative resolution of a combined veterinary formulation by ANN and PCA-ANN methods

The simultaneous spectral prediction of levamisole (LEV) and triclabendazole (TRI) in combined veterinary formulation was performed by the new chemometric methods, artificial neural network (ANN) and principal component analysis-artificial neural network (PCA-ANN). Despite the overlapping spectra of LEV and TRI in the same wavelength region, the proposed methods do not use any separation procedure for the analysis of the related compounds. Good precision and accuracy were observed for the applications of the proposed artificial neural network models to an independent binary mixture set consisting of the active compounds. These methods were successfully applied for the chemometric quantitation of a veterinary formulation of LEV and TRI.

Comparison of Calibration Models Based on Near Infrared Spectroscopy Data for the Determination of Plant Oil Properties

The aim of this study was to compare the prediction efficiency of different type of linear calibration models using near infrared (NIR) absorbance spectral data of vegetable oils. The applied model types were the PCA-MLR (principal component analysis-multiple linear regression), the PLS (partial least squares regression), the PCA-ANN (principal component analysis-artificial neural network) and the GA-ANN (genetic algorithm-artificial neural network). The calibrations were executed on the models for the determination of the concentration of oleic acid of vegetable oils and the performances of the different models were determined using external validation. During external validation the built models were tested with vegetable oil samples of which oleic acid content was known and was not included in the calibration sample set. The comparison of the models was executed on the basis of the accuracy of the prediction.

PREDICTING Ms TEMPERATURE APPLYING PRINCIPAL COMPONENT ANALYSIS-ARTIFICIAL NEURAL NETWORKS

The principal component analysis-artificial neural network (PCA-ANN) model was developed to predict martensite transformation start temperature ( Ms ) of steels. Training samples were processed by principal component analysis and the number of input variables was reduced from 6 to 4, then the scores of principal components were used to establish new sample database to train the ANN model. Ms of steels were predicted by the PCA-ANN model. The predicted and measured Ms distribute along the 0-45° diagonal in the scatter diagram and the statistical errors are MSE-16.0256, MSRE-4.49% and VOF-1.97790 respectively. Comparing the prediction results of different models it is shown that the accuracy of the PCA-ANN model was the highest, which indicated that the principal component analysis was helpful to improve the prediction accuracy of ANN model.