Simple Artificial Neural Network Research Articles

Simultaneous identification of groundwater pollution source location and release concentration using Artificial Neural Network

Groundwater pollution source identification problem plays an important role in designing the groundwater remediation measures. This study aims at identifying the groundwater pollutant source location and its strength given a set of concentration breakthrough curves at different locations. Feed forward three layered artificial neural network (ANN) has been used to identify these two source parameters: location, i.e., the distance of source from the observation point and strength, i.e., the release concentration. A simplified ANN architecture is achieved by presenting the available concentration breakthrough curves in a specific manner as input to the model. The database of 1750 patterns for training and testing of the model is generated employing analytical solution of one dimensional steady flow and transient contaminant transport in a homogeneous aquifer. The performance of the ANN model was evaluated using standard statistical methods. A network with architecture 22-16-2 was found to be optimum and resulted in a normalized root mean square error (NRMSE) of 0.014 for identifying the source location and NRMSE of 0.044 for identifying the source release concentration. The results show that ANN can be a very efficient tool for locating pollution sources and estimating the release concentration at source. A good ANN model performance was obtained even with a simple architecture and with a small number of input variables.

Forecasting Rice Production of Bangladesh Using ARIMA and Artificial Neural Network Models

Forecasting behavior of Econometric and Machine Learning models has recently attracted much attention in the research sector. In this study an attempt has been made to compare the forecasting behavior of Autoregressive Integrated Moving Average (ARIMA) and Artificial Neural Network (ANN) using univariate time series data of annual rice production (1972 to 2013) of Bangladesh. Here, suitable ARIMA has been chosen from several selected ARIMA models with the help of AIC and BIC values. A simple ANN model using backpropagation algorithm with appropriate number of nodes or neurons in a single hidden layer, adjustable threshold value and learning rate, has been constructed. Based on the RMSE, MAE and MAPE values, the results showed that the estimated error of ANN is much higher than the estimated error of chosen ARIMA. So, according to this study, it can be said that the ARIMA model is better than ANN model for forecasting the rice production in Bangladesh.
 Dhaka Univ. J. Sci. 68(2): 143-147, 2020 (July)

Valorization of Vegetable Food Waste and By-Products Through Fermentation Processes.

There is a general interest in finding new ways of valorizing fruit and vegetable processing by-products. With this aim, applications of industrial fermentation to improve nutritional value, or to produce biologically active compounds, have been developed. In this sense, the fermentation of a wide variety of by-products including rice, barley, soya, citrus, and milling by-products has been reported. This minireview gives an overview of recent fermentation-based valorization strategies developed in the last 2 years. To aid the designing of new bioprocesses of industrial interest, this minireview also provides a detailed comparison of the fermentation conditions needed to produce specific bioactive compounds through a simple artificial neural network model. Different applications reported have been focused on increasing the nutritional value of vegetable by-products, while several lactic acid bacteria and Penicillium species have been used to produce high purity lactic acid. Bacteria and fungi like Bacillus subtilis, Rhizopus oligosporus, or Fusarium flocciferum may be used to efficiently produce protein extracts with high biological value and a wide variety of functional carbohydrates and glycosidases have been produced employing Aspergillus, Yarrowia, and Trichoderma species. Fermentative patterns summarized may guide the production of functional ingredients for novel food formulation and the development of low-cost bioprocesses leading to a transition toward a bioeconomy model.

Predicting particle trajectories in oceanic flows using artificial neural networks

Predicting ocean transport has many practical applications ranging from search and rescue operations to predicting the spread of oil, debris, and biogeochemical tracers, yet trajectory prediction remains a challenge for existing ocean modeling techniques. General circulation models require high resolution observational data in order to be properly initialized, but these data do not exist for the ocean. Statistical models are difficult to tune with existing data and are often too simple to accurately encapsulate turbulent flows. Here we investigate a data-driven approach to ocean transport prediction wherein the goal is to first learn from available data instead of prescribed laws of physics and then apply this information to new data. More specifically, we explore whether simple artificial neural networks (ANNs) are capable of learning to predict 2D particle trajectories using only previous velocity observations. ANNs are trained in two ways: first, a so-called “one-to-one ANN” uses a particle’s most recently observed velocity to predict its velocity six hours later, and second, a “time series ANN” uses the past 24 hours’ worth of velocity observations to predict the next 24 h. We present a proof-of-concept considering particles in a hierarchy of simulated flow regimes ranging from uniform, steady flow to more complex scenarios with interacting scales of motion and then substantiate our approach on trajectories in modeled flows generated by a high-resolution Hybrid Coordinate Ocean Model for a mesoscale eddy in the northern Gulf of Mexico. We also assess ANN sensitivity to the prediction window over which forecasts are made, the number of training particles used, and the size of the network. ANNs successfully predict 24 h trajectories within the temporal bounds of the training data with forecast errors around half those of both rudimentary persistence and classical ARIMA models. Predicting beyond the domain of the training data leads to forecast errors comparable to ARIMA models. Our results suggest that ANNs offer promising potential as a data-driven approach to forecasting material transport in the ocean.

Gamma ray log generation from drilling parameters using deep learning

Lithology identification plays a vital role in defining the petroleum reservoir. Although well logging represents the traditional means of obtaining petrophysical data for lithology identification, there could be cases where logging while drilling instruments may fail during drilling. This paper presents an approach that utilizes drilling parameters obtained from mud logging and measurement while drilling (MWD) for real-time prediction of gamma ray log which is used as a lithology identifier. In this work, several machine learning methodologies, such as simple recurrent neural network (RNN), long short-term memory recurrent neural network (LSTM-RNN), temporal convolution network (TCN), gated recurrent unit (GRU) network, nonlinear autoregressive network with exogenous inputs (NARX) and simple artificial neural network (ANN) were tested for their ability to capture the relationship between hydro-mechanical specific energy (computed from drilling parameters) and gamma ray log. A recently drilled exploration gas well in the tertiary deltaic system of the Niger delta basin was used as the case study. Based on the field data, the results show the TCN and simple RNN performed best. The receptive field of the TCN played a significant role in its performance, and therefore, the LSTM-RNN can be made to perform comparable to that of TCN/simple RNN if the LSTM-RNN is manually made to work with an optimal window of input data points for each output data point. The size and nature of the data (volume, velocity, variety, and veracity) is likely a significant factor in the performance of the machine learning methodologies. Thus, it is recommended to focus on obtaining the best receptive field of the data during the machine learning development phase.

Topographic property of backpropagation artificial neural network: From human functional connectivity network to artificial neural network

The human brain network is one of the most complex data processing centers so far and shows several topographic features such as small worldness, modularity, and high efficiency. Inspired by the human brain network, the artificial neural network (ANN) method was proposed and has become an important engineering tool for pattern recognition. A recent study suggested the similarity between the human brain network and ANN. Would the ANN show topographic features that are similar to those of the human brain network? In the present study, we first constructed a simple backpropagation ANN (BP-ANN) with one hidden layer, which is one of the simplest, representative ANN, based on CIFAR10 and CIFAR100 datasets. Then, we explored the topographic features of a network with artificial neurons. We found that artificial neurons of the BP-ANN showed a high local efficiency and modularity similar to the human brain network. However, the global efficiency of artificial neurons of the BP-ANN was comparable with those of a random network. Our results showed that the best model (high accuracy and no overfitting) of a large-scale BP-ANN during epochs of training showed the lowest local and global efficiency. These results might imply that after epochs of training, neurons of the BP-ANN form local highly intra-connected functional modules but lack cooperation at the global level. The topography structure of the BP-ANN also shows the potential to become an early-stop marker during training.

A SIMPLE ARTIFICIAL NEURAL NETWORK FOR FIRE DETECTION USING LANDSAT-8 DATA

Abstract. Fixed threshold models have been widely used in active fire detection products. However, its accuracy is limited due to the complexity of setting up thresholds. Artificial neural network (ANN) is capable of learning from data and can decide weights automatically. Given enough data, an ANN model is able to optimize itself and quickly find an optimal solution. In this work, a simple ANN model is implemented to classify fire pixels from Landsat-8 data. Experimental results show that our ANN model effectively achieves fire detection and performs better than fixed threshold model in certain circumstances.

Significance of artificial neural network analytical models in materials’ performance prediction

In materials science, performance prediction of materials plays an important role in improving the quality of materials as well as preventing serious damage to the environment and threat to public safety. Traditional regression analysis models in materials science are not yet perfect, limited to capture nonlinearities of data and time-consumption for prediction, and have a poor ability to handle a large amount of data. This leads to a demand for analyses of materials data using novel computer science methods. In recent years, artificial neural networks (ANNs) are increasingly performing as a strong tool to establish the relationships among data and being successfully applied in materials science due to their generalization ability, noise tolerance and fault tolerance. In this paper, some typical ANN applications for predicting various properties (corrosion, structural, tribological and so on) of different materials serving multiple environments (atmosphere, stress, weld and so on) are reviewed. It highlights the significance of ANN in materials-related problems in separate sections arranged by the level of simplicity, ranging from simple ANN models alone to more complicated ANN models along with the hybrid use of other computing and input-ranking methods, and the trend of ANN in the context of materials science with some limitations.

Application of Artificial Neural Networks for Natural Gas Consumption Forecasting

The present research study explores three types of neural network approaches for forecasting natural gas consumption in fifteen cities throughout Greece; a simple perceptron artificial neural network (ANN), a state-of-the-art Long Short-Term Memory (LSTM), and the proposed deep neural network (DNN). In this research paper, a DNN implementation is proposed where variables related to social aspects are introduced as inputs. These qualitative factors along with a deeper, more complex architecture are utilized for improving the forecasting ability of the proposed approach. A comparative analysis is conducted between the proposed DNN, the simple ANN, and the advantageous LSTM, with the results offering a deeper understanding the characteristics of Greek cities and the habitual patterns of their residents. The proposed implementation shows efficacy on forecasting daily values of energy consumption for up to four years. For the evaluation of the proposed approach, a real-life dataset for natural gas prediction was used. A detailed discussion is provided on the performance of the implemented approaches, the ANN and the LSTM, that are characterized as particularly accurate and effective in the literature, and the proposed DNN with the inclusion of the qualitative variables that govern human behavior, which outperforms them.

Artificial neural networks improve early outcome prediction and risk classification in out-of-hospital cardiac arrest patients admitted to intensive care

BackgroundPre-hospital circumstances, cardiac arrest characteristics, comorbidities and clinical status on admission are strongly associated with outcome after out-of-hospital cardiac arrest (OHCA). Early prediction of outcome may inform prognosis, tailor therapy and help in interpreting the intervention effect in heterogenous clinical trials. This study aimed to create a model for early prediction of outcome by artificial neural networks (ANN) and use this model to investigate intervention effects on classes of illness severity in cardiac arrest patients treated with targeted temperature management (TTM).MethodsUsing the cohort of the TTM trial, we performed a post hoc analysis of 932 unconscious patients from 36 centres with OHCA of a presumed cardiac cause. The patient outcome was the functional outcome, including survival at 180 days follow-up using a dichotomised Cerebral Performance Category (CPC) scale with good functional outcome defined as CPC 1–2 and poor functional outcome defined as CPC 3–5. Outcome prediction and severity class assignment were performed using a supervised machine learning model based on ANN.ResultsThe outcome was predicted with an area under the receiver operating characteristic curve (AUC) of 0.891 using 54 clinical variables available on admission to hospital, categorised as background, pre-hospital and admission data. Corresponding models using background, pre-hospital or admission variables separately had inferior prediction performance. When comparing the ANN model with a logistic regression-based model on the same cohort, the ANN model performed significantly better (p = 0.029). A simplified ANN model showed promising performance with an AUC above 0.852 when using three variables only: age, time to ROSC and first monitored rhythm. The ANN-stratified analyses showed similar intervention effect of TTM to 33 °C or 36 °C in predefined classes with different risk of a poor outcome.ConclusionA supervised machine learning model using ANN predicted neurological recovery, including survival excellently, and outperformed a conventional model based on logistic regression. Among the data available at the time of hospitalisation, factors related to the pre-hospital setting carried most information. ANN may be used to stratify a heterogenous trial population in risk classes and help determine intervention effects across subgroups.

Detection of Parkinson’s Disease from gait using Neighborhood Representation Local Binary Patterns

Parkinson's Disease (PD) is a neurodegenerative disease that affects millions of people around the world. Diagnostics tools based on the clinical symptoms have been developed by the scientific community mostly in the last decade. This study proposes a new method of PD detection from gait signals, using artificial neural networks and a novel technique framework called Neighborhood Representation Local Binary Pattern (NR-LBP). Vertical Ground Reaction Force (VGRF) readings are preprocessed and transformed using several methods within the proposed framework. Statistical features are extracted from the transformed data, and the Student's t-test test is used to create different feature sets. A simple artificial neural network is trained over these features to detect PD, and its performance is evaluated using different metrics. Classification accuracy of 98.3% and Matthews Correlation Coefficient of 0.959 are obtained, indicating high-performance classification. Based on these performance measures, the proposed NR-LBP algorithm is compared to the regular LBP algorithm and found to be contributing positively to classification performance when various types of transformations are used in combination.

Incremental Dilations Using CNN for Brain Tumor Classification

Brain tumor classification is a challenging task in the field of medical image processing. Technology has now enabled medical doctors to have additional aid for diagnosis. We aim to classify brain tumors using MRI images, which were collected from anonymous patients and artificial brain simulators. In this article, we carry out a comparative study between Simple Artificial Neural Networks with dropout, Basic Convolutional Neural Networks (CNN), and Dilated Convolutional Neural Networks. The experimental results shed light on the high classification performance (accuracy 97%) of Dilated CNN. On the other hand, Dilated CNN suffers from the gridding phenomenon. An incremental, even number dilation rate takes advantage of the reduced computational overhead and also overcomes the adverse effects of gridding. Comparative analysis between different combinations of dilation rates for the different convolution layers, help validate the results. The computational overhead in terms of efficiency for training the model to reach an acceptable threshold accuracy of 90% is another parameter to compare the model performance.

Gender identification for Egyptian Arabic dialect in twitter using deep learning models

Although the number of Arabic language writers in social media is increasing, the research work targeting Author Profiling (AP) is at the initial development phase. This paper investigates Gender Identification (GI) (male or female) of authors posting Egyptian dialect tweets using Neural Networks (NN) models. Various architectures of NN are explored with extensive parameters’ selection such as simple Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Long–Short Term Memory (LSTM), Convolutional Bidirectional Long-Short Term Memory (C-Bi-LSTM) and Convolutional Bidirectional Gated Recurrent Units (C-Bi-GRU) NN which is tuned for the GI problem at hand. The best acquired GI accuracy using C-Bi-GRU multichannel model is 91.37%. It is worth noting that the presence of the bidirectional layer as well as the convolutional layer in the NN models has significantly enhanced the GI accuracy.

Fractal structure in the S&P500: A correlation-based threshold network approach

• A correlation-based threshold network to examine the network fractality is proposed. • A self-similarity aspects of the S&P500 network is revealed. • A strong effective repulsion phenomenon is detected. • Different growth patterns are observed for various cases of S&P500 network. • Proposed measures improve the prediction performance for long-term future. This research aims to analyze the S&P500 network, one of the representatives of the global financial market, based on its network fractality. The research is conducted in the following steps. At first, we propose the concept of a correlation-based threshold network based on minimum spanning tree. Secondly, we investigate the fractal dimension of threshold networks and propose suitable fractal dimension measures. Lastly, we analyze the S&P500 network based on the proposed measures and utilize them in the market prediction. Based on the results, we discover the self-similarity characteristic of the S&P500 network, where a strong effective repulsion phenomenon is detected. Furthermore, we observe the different growth patterns of S&P500 network for different combinations of fractal conditions defined by the proposed measures. Then, we utilize the measures in the prediction of the cumulative log-return of S&P500 index via a simple artificial neural network and detect the improvement of prediction performance in the long-term development of the market.

Combined use of principal component analysis and artificial neural network approach to improve estimates of PM2.5 personal exposure: A case study on older adults

Accurate exposure estimate of the air pollutant PM2.5 is required to evaluate its health impacts in epidemiological studies, due to its adverse effects on human's respiratory and cardiovascular systems. However, traditional personal sampling is time and cost consuming. Thus, modeling techniques are needed to accurately predict the personal exposure level to PM2.5. In this study, a total of 117 older adults over 60 were recruited in Tianjin, a heavily polluted city in northern China, for indoor, outdoor and personal PM2.5 sampling. Eighteen variables which may increase the exposure level of older adults were recorded for artificial neural network (ANN) simulation. Four modeling techniques, including time-integrated activity modeling, Monte Carlo simulation, ANN modeling, and combined use of principal component analysis (PCA) and ANN model, were used to evaluate their ability for predicting real exposure values of PM2.5. The results of traditional time-weighted activity modeling showed the lowest correlation with measured values with R2 of 0.57 and 0.42 in winter and summer, respectively. For Monte Carlo simulation, high correlation was obtained (R2 of 0.93 and 0.92 in winter and summer, respectively) between percentiles of the predicted and the real exposure values. Compared with the simple ANN models, the combined use of PCA and ANN produced the most accurate results with R2 of 0.99 and RMSE lower than 15. Since the information of the input variables for the PCA-ANN model can be obtained from the questionnaire and fixed air quality monitoring sites, this technique shows a great potential in predicting personal exposure level to the air pollutant because no additional concentration measurement is needed.

Rational Band Engineering of an Organic Double Heterojunction for Artificial Synaptic Devices with Enhanced State Retention and Linear Update of Synaptic Weight.

Herein, we propose an organic double heterojunction to enable a nonvolatile step modulation of the conductance of an artificial synapse; the double heterojunction is composed of N,N'-dioctyl-3,4,9,10-perylene tetracarboxylic diimide (PTCDI-C8), copper phthalocyanine (CuPc), and para-sexiphenyl (p-6P). The carrier confinement in the CuPc region present in the double-heterojunction structure enabled the nonvolatile modulation of the postsynaptic current. The proposed organic synapse exhibited an excellent conductance change, characteristic with a nonlinearity (NL) value below 0.01 in the long-term potentiation (LTP) region. Furthermore, the NL value for long-term depression (LTD) could be reduced effectively from 45 to 3.5 by a pulse modulation technique. A simple artificial neural network (ANN) was theoretically designed using the LTP/LTD characteristic curves of such organic synapses, and then, learning and recognition tasks were performed using Modified National Institute of Standards and Technology digit images. A four-amplitude weight update method enabled considerable enhancement of the recognition rate from 53 to 70%. Although the designed ANN was based on a single-layer perceptron model, a high maximum accuracy of 75% was achieved. These newly studied techniques for synaptic devices are expected to open up new possibilities for the realization of artificial synapses based on organic double heterojunctions.

Developing a novel artificial intelligence model to estimate the capital cost of mining projects using deep neural network-based ant colony optimization algorithm

This study aims to propose a novel artificial intelligence model for forecasting the capital cost (CC) of open-pit mining projects with high accuracy. It is a unique combination of a deep neural network (DNN) and ant colony optimization (ACO) algorithm, abbreviated as ACO-DNN. In this model, MineAP (annual mine production), SR (stripping ratio), MillAP (annual production of the mill), RMG (reserve mean grade), and LOM (life of mine) were used to consider the CC of open-pit mining projects. A series of simple and complex artificial neural networks (ANN) was developed for forecasting CC of 74 copper mining projects herein. Subsequently, the ACO algorithm has been applied to optimize the developed ANN and DNN models to improve the accuracy of them. Finally, an optimal hybrid model was defined (i.e., ACO-DNN 5-25-20-18-15-1) with superior performance than other models (i.e., RMSE of 130.988, R2 of 0.991, MAE of 115.274, MAPE of 0.072, and VAF of 99.052). The findings of this study showed that the DNN models could predict the CC for open-pit mining projects with more accuracy than those of the simple ANN models. In particular, the ACO algorithm played an essential role in improving the accuracy of forecasting models. Also, MineAP, MillAP, SR, and LOM have been confirmed as critical parameters that affect the accuracy of the selected model in forecasting the CC of open-pit mining projects, especially MineAP. In conclusion, this study offers a useful tool to improve resource policies of mining projects, especially copper mining projects.

Migrating from partial least squares discriminant analysis to artificial neural networks: a comparison of functionally equivalent visualisation and feature contribution tools using jupyter notebooks

IntroductionMetabolomics data is commonly modelled multivariately using partial least squares discriminant analysis (PLS-DA). Its success is primarily due to ease of interpretation, through projection to latent structures, and transparent assessment of feature importance using regression coefficients and Variable Importance in Projection scores. In recent years several non-linear machine learning (ML) methods have grown in popularity but with limited uptake essentially due to convoluted optimisation and interpretation. Artificial neural networks (ANNs) are a non-linear projection-based ML method that share a structural equivalence with PLS, and as such should be amenable to equivalent optimisation and interpretation methods.ObjectivesWe hypothesise that standardised optimisation, visualisation, evaluation and statistical inference techniques commonly used by metabolomics researchers for PLS-DA can be migrated to a non-linear, single hidden layer, ANN.MethodsWe compared a standardised optimisation, visualisation, evaluation and statistical inference techniques workflow for PLS with the proposed ANN workflow. Both workflows were implemented in the Python programming language. All code and results have been made publicly available as Jupyter notebooks on GitHub.ResultsThe migration of the PLS workflow to a non-linear, single hidden layer, ANN was successful. There was a similarity in significant metabolites determined using PLS model coefficients and ANN Connection Weight Approach.ConclusionWe have shown that it is possible to migrate the standardised PLS-DA workflow to simple non-linear ANNs. This result opens the door for more widespread use and to the investigation of transparent interpretation of more complex ANN architectures.

Transmembrane Alpha Helix Prediction Using Neural Networks

This paper presents a simple artificial neural network which classifies proteins into two classes from their sequences alone: the alpha helix transmembrane protein class and the non-alpha helix transmembrane protein class. The network described here has a simple feed-forward topology and a limited number of neurons which makes it very fast.

Use of Empirical Mode Decomposition in Improving Neural Network Forecasting of Paddy Price

Since rice is a staple food in Malaysia, its price fluctuations pose risks to the producers, suppliers and consumers. Hence, an accurate prediction of paddy price is essential to aid the planning and decision-making in related organizations. The artificial neural network (ANN) has been widely used as a promising method for time series forecasting. In this paper, the effectiveness of integrating empirical mode decomposition (EMD) into an ANN model to forecast paddy price is investigated. The hybrid method is applied on a series of monthly paddy prices fromFebruary 1999 up toMay 2018 as recorded in the Malaysian Ringgit (MYR) per metric tons. The performance of the simple ANN model and the EMD-ANN model was measured and compared based on their root mean squared Error (RMSE), mean absolute error (MAE) and mean percentage error (MPE). This study finds that the integration of EMD into the neural network model improves the forecasting capabilities. The use of EMD in the ANN model made the forecast errors reduced significantly, and the RMSE was reduced by 0.012, MAE by 0.0002 and MPE by 0.0448.