Ensemble Artificial Neural Network and Support Vector Machine Based Parameter Evaluation in Wireless Sensor Networks

The authors provide a method based on ensembles of artificial neural networks (ANNs) that, fed with supervised training data, are able to estimate rotational speed of the machine under investigation. Ensembles of various ANNs trained using different algorithms provide reliability of the method and protection from unfortunate initial weights distribution. The method is validated on the basis of data gathered in large amount of experiments performed in epicyclic gearbox test bed

In this paper, operational and complexity analysis are investigated for a proposed model of ensemble Artificial Neural Networks (ANN) multiple classifiers. The main idea to this is to employ more classifiers to obtain a more accurate prediction as well as to enhance the classification capabilities in case of larger data. The classification result analyzed between a single classifier and multiple classifiers followed by the estimates of upper bounds of converged functional error with the partitioning of the benchmark dataset. The estimates derived using the Apriori method shows that the proposed ensemble ANN algorithm with a different approach is feasible where such problems with a high number of inputs and classes can be solved with time complexity of O(n^k ) for some k, which is a type of polynomial. This result is in line with the significant performance achieved by the diversity rule applied with the use of reordering technique. As conclusion, an ensemble heterogeneous ANN classifier is practical and relevant to theoretical and experimental of combiners for the ensemble ANN classifier systems for a large dataset.

To develop a QSAR model, based on calculated molecular descriptors and an Artificial Neural Networks Ensemble (ANNE), for the estimation of rat tissue-to-blood partition coefficients (Kt:b), as well as the assessment of the applicability domain of the model and its utility in predicting the drug distribution in humans. A total of 1460 individual Kt:b values (75% train and 25% validation), obtained in 13 different rat tissues were collected in the literature. A correlation between simple molecular descriptors for lipophilicity, ionization, size and hydrogen bonding capacity and Kt:b data was attempted by using an ANNE. Similar statistics were observed between the train and validation group of data with correlations, between the observed values and the predicted average ANNE values, of 0.909 and 0.896, respectively. A degradation of the correlations was observed for predicted values with high uncertainty, as judged by the standard deviations of the ANNE outputs. This was further observed when using the ANNE Kt:b values in a Physiologically based pharmacokinetic (PBPK) model for predicting the Human Volume of distribution of another 532 drugs. This model (available as a MS Excel® workbook in the Supporting material of this article) may be a valuable tool for prediction and simulation in early drug development, allowing the in silico estimation of rat Kt:b values for PBPK purposes and also indicating its applicability domain.

A simple, efficient, and fast method based on in-syringe dispersive liquid–liquid microextraction (IS-DLLME) for preconcentration of trace amounts of palladium from aqueous samples was developed. After complexation with 5-(4-dimethylaminobenzylidene)rhodanine, Pd was extracted into benzyl alcohol before its measurement with UV–Vis spectrophotometer, equipped with cubic millimeter cells. Thereafter, a comparative study between single best artificial neural network (SB-NN) and neural network ensemble (NNE) was performed to find the best mathematical model for palladium extraction process to simulate IS-DLLME. Two NNE models were built, one without pruning (NNE-WP) the ensemble members and another with pruning using genetic algorithm (NNE-GA). The predictive and generalization ability of SB-NN, NNE-WP, and NNE-GA was compared based on 20 runs. The average % error for SB-NN, NNE-WP, and NNE-GA models was 0.234, 0.146, and 0.115 and the correlation coefficient was 0.902, 0.948, and 0.973, respectively; indicating superiority of NNE approaches specially NNE-GA in capturing the non-linear behavior of the system.

Terra and Aqua, 2 satellites launched by the NASA-centered international Earth Observing System project, house MODIS (Moderate Resolution Imaging Spectroradiometer) sensors. Moderate resolution remote sensing allows the quantifying of land surface type and extent, which can be used to monitor changes in land cover and land use for extended periods of time. In this paper, we propose applying an ensemble technique, based on fault masking among individual classifier for N-version programming. We create an N-version programming ensemble of artificial neural networks and use the majority voting result to predict land surface cover from MODIS data. We show that an N-version programming ensemble of neural networks greatly improves the classification error rate of land cover type.

PhotoVoltaic (PV) plants can provide important economic and environmental benefits to electric systems. On the other hand, the variability of the solar source leads to technical challenges in grid management as PV penetration rates increase continuously. For this reason, PV power forecasting represents a crucial tool for uncertainty management to ensure system stability. In this paper, a novel hybrid methodology for the PV forecasting is presented. The proposed approach can exploit clear-sky models or an ensemble of artificial neural networks, according to day-ahead weather forecast. In particular, the selection among these techniques is performed through a decision tree approach, which is designed to choose the best method among those aforementioned. The presented methodology has been validated on a real PV plant with very promising results.

Solar power generation is highly uncertain and intermittent due to its strong dependence on climate and meteorological factors. In this study, we have proposed a bagging ensemble of artificial neural network (BagANN) model with weighted averaging optimized using grey wolf optimization (GWO) and investigate the impact of climatic and meteorological factors on the output power prediction of the solar power plant. Initially, we have preprocessed the data by applying feature engineering, normalization techniques, and appropriate features are selected for power prediction. The hyperparameters of the BagANN model are tuned and optimized using the GWO technique to improve the models predictive accuracy. Experiments are conducted to evaluate the generalization performance of the developed model using five-fold and 10-fold cross-validation techniques. The performance results are assessed using RMSE, coefficient of determination ( R 2 ), adjusted ( R 2 ), and the MAPE. Results show that the BagANN with weighted averaging has a higher level of precision with R 2 of 98.5% predictive accuracy, has less prediction errors with 0.1263 MSE and 0.0267 MAPE than the other developed models. From the result, it is evident that the BagANN model has significant improvements compared to other models available in the literature in terms of robustness of the prediction and predictive accuracy.

Using wireless sensor networks (WSNs) in delay and reliability critical environments is highly desired due to their unique advantages such as low cost, ease of deployment, and redundancy. However, successful deployment of resource-limited WSNs in such applications requires strict quality-of-service (QoS) provisioning techniques to meet the desired latency and reliability targets of those applications. Implementation of QoS techniques in WSNs is significantly challenging since WSNs have been initially devised for low-data-rate non-real-time applications. Therefore, WSN designers and developers resort to different cross-layer interaction and optimization techniques to provision QoS in WSNs. In this paper, we present a survey on the state of the art of cross-layer QoS approaches in wireless terrestrial sensor networks to achieve delay and reliability bounds in critical applications. Our paper provides a unique classification of cross-layer QoS approaches in WSNs that allows surveying a large amount of studies with utmost clarity. Furthermore, we highlight the main challenges of implementing QoS protocols in WSNs and present an overview of QoS-aware WSN applications.

Inhaling particulate matter such as PM2.5 can have a hazardous impact on the human health. In order to predict the PM2.5 concentration, Artificial Neural Networks trained with conjugate gradient descent such as Multi Layer Perceptron (MLP), cascade forward neural network, Elman neural network, Radial Basis Function (RBF) neural network and Non-linear Autoregressive model with exogenous input (NARX) along with regression models such as Multiple Linear Regression (MLR) consisting of batch gradient descent, stochastic gradient descent, mini-batch gradient descent and conjugate gradient descent algorithms and Support Vector Regression (SVR) were implemented. In these models, the concentration of PM2.5 was the dependent variable and the data related to concentrations of PM2.5, SO2, O3 and meteorological data including average Maximum Temperature (MAX T), daily wind speed (WS) for the years 2010–2016 in Houston and New York were the independent variables. For the final forecast, several ensemble models of individual neural network predictors and individual regression predictors have been presented.

A study was initiated to predict water quality index (WQI) using artificial neural networks (ANNs) with respect to the concentrations of 16 groundwater quality variables collected from 47 wells and springs in Andimeshk during 2006–2013 by the Iran’s Ministry of Energy. Such a prediction has the potential to reduce the computation time and effort and the possibility of error in the calculations. For this purpose, three ANN’s algorithms including ANNs with early stopping, Ensemble of ANNs and ANNs with Bayesian regularization were utilized. The application of these algorithms for this purpose is the first study in its type in Iran. Comparison among the performance of different methods for WQI prediction shows that the minimum generalization ability has been obtained for the Bayesian regularization method (MSE = 7.71) and Ensemble averaging method (MSE = 9.25), respectively and these methods showed the minimum over-fitting problem compared with that of early stopping method. The correlation coefficients between the predicted and observed values of WQI were 0.94 and 0.77 for the test and training data sets, respectively indicating the successful prediction of WQI by ANNs through Bayesian regularization algorithm. A sensitivity analysis was implemented to show the importance of each parameter in the prediction of WQI during ANN’s modeling and showed that parameters like Phosphate and Fe are the most influential parameters in the prediction of WQI.

Artificial neural networks have been successfully applied to a variety of machine learning tasks, including image recognition, semantic segmentation, and machine translation. However, few studies fully investigated ensembles of artificial neural networks. In this work, we investigated multiple widely used ensemble methods, including unweighted averaging, majority voting, the Bayes Optimal Classifier, and the (discrete) Super Learner, for image recognition tasks, with deep neural networks as candidate algorithms. We designed several experiments, with the candidate algorithms being the same network structure with different model checkpoints within a single training process, networks with same structure but trained multiple times stochastically, and networks with different structure. In addition, we further studied the overconfidence phenomenon of the neural networks, as well as its impact on the ensemble methods. Across all of our experiments, the Super Learner achieved best performance among all the ensemble methods in this study.

An ensemble neural network atmospheric correction for Sentinel-3 OLCI over coastal waters providing inherent model uncertainty estimation and sensor noise propagation

Electrical consumption forecasting in sports venues: A proposed approach based on neural networks and ARIMAX Models

We examine a classification problem in which seismic waveforms of natural earthquakes are to be distinguished from waveforms of man-made explosions. We present an integrated classification machine (ICM), which is a hierarchy of artificial neural networks (ANNs) that are trained to classify the seismic waveforms. In order to maximize the gain of combining the multiple ANNs, we suggest construction of a redundant classification environment (RCE) that consists of several whose expertise depends on the different input representations to which they are exposed. In the proposed scheme, the experts are ensembles of ANN, trained on different bootstrap replicas. We use various network architectures, different time-frequency decompositions of the seismic waveforms, and various smoothing levels in order to achieve an RCE. A confidence measure for the ensemble's classification is defined based on the agreement (variance) within the ensembles, and an algorithm for a nonlinear integration of the ensembles using this measure is presented. An implementation on a data set of 380 seismic events is described, where the proposed ICM had classified correctly 92% of the testing signals. The comparison we made with classical methods indicates that combining a collection of ensembles of ANNs can be used to handle complex high dimensional classification problems.

Over the past couple of years, the share of wind power in electrical power system has increased considerably. Because of the irregular characteristics of wind, the power generated by the wind turbines fluctuates continuously. The unstable nature of the wind power thus poses a serious challenge in power distribution systems. For reliable power distribution, wind power prediction system has become an essential component in power distribution systems. In this Paper, a wind power forecasting strategy composed of Artificial Neural Networks (ANN) and Genetic Programming (GP) is proposed. Five neural networks each having different structure and different learning algorithm were used as base regressors. Then the prediction of these neural networks along with the original data is used as input for GP based ensemble predictor. The proposed wind power forecasting strategy is applied to the data from five wind farms located in same region of Europe. Numerical results and comparison with existing wind power forecasting strategies demonstrates the efficiency of the proposed strategy.