Search

As entertainment tools, computer games are important phenomena in the world, which are considered as a popular medium, an effective educational solution and a considerable economy resource. In this paper, Multi-Layer perceptron (MLP) neural network was used to detect human behavior pattern in rock, paper, scissors game. The similarity of artificial neural networks (ANNs) to the human brain is the main motivation of this study. MATLAB software was used to implement the network code. These codes consisted of two phases: 1) training the ANN to learn the human behavioral pattern considering forty games. 2) real play against a human by doing ten games. After the implementation of the network, its effectiveness in detecting human behavioral patterns was investigated. The network was tested on 40 people (20 women and 20 men). Each player played with the target network in three stages. The results of this study showed that the win percentage of computers with MLP neural network was 57.5% for men and 60.8% for women. While the percentage of the computer without neural networks and with random selections in 60 games was 52.5% for men and 42.5% for women

The gallium-nitride (GaN) high electron-mobility transistor (HEMT) technology has emerged as an attractive candidate for high-frequency, high-power, and high-temperature applications due to the unique physical characteristics of the GaN material. Over the years, much effort has been spent on measurement-based modeling since accurate models are essential for allowing the use of this advanced transistor technology at its best. The present analysis is focused on the modeling of the scattering (S-) parameter measurements for a 0.25 μm GaN HEMT on silicon carbide (SiC) substrate at extreme operating conditions: a large gate width (i.e., the transistor is based on an interdigitated layout consisting of ten fingers, each with a length of 150 μm, resulting in a total gate periphery of 1.5 mm), a high ambient temperature (i.e., from 35 °C up to 200 °C with a step of 55 °C), a high dissipated power (i.e., 5.1 W at 35 °C), and a high frequency in the millimeter-wave range (i.e., from 200 MHz up to 65 GHz with a step of 200 MHz). Three different modeling approaches are investigated: the equivalent-circuit model, artificial neural networks (ANNs), and gated recurrent units (GRUs). As is shown, each modeling approach has its pros and cons that need to be considered, depending on the target performance and their specifications. This implies that an appropriate selection of the transistor modeling approach should be based on discerning and prioritizing the key features that are indeed the most important for a given application.

In this study, we propose the use of artificial synaptic transistors with coplanar-gate structures fabricated on paper substrates comprising biocompatible and low-cost potato-starch electrolyte and indium–gallium–zinc oxide (IGZO) channels. The electrical double layer (EDL) gating effect of potato-starch electrolytes enabled the emulation of biological synaptic plasticity. Frequency dependence measurements of capacitance using a metal-insulator-metal capacitor configuration showed a 1.27 μF/cm2 at a frequency of 10 Hz. Therefore, strong capacitive coupling was confirmed within the potato-starch electrolyte/IGZO channel interface owing to EDL formation because of internal proton migration. An electrical characteristics evaluation of the potato-starch EDL transistors through transfer and output curve resulted in counterclockwise hysteresis caused by proton migration in the electrolyte; the hysteresis window linearly increased with maximum gate voltage. A synaptic functionality evaluation with single-spike excitatory post-synaptic current (EPSC), paired-pulse facilitation (PPF), and multi-spike EPSC resulted in mimicking short-term synaptic plasticity and signal transmission in the biological neural network. Further, channel conductance modulation by repetitive presynaptic stimuli, comprising potentiation and depression pulses, enabled stable modulation of synaptic weights, thereby validating the long-term plasticity. Finally, recognition simulations on the Modified National Institute of Standards and Technology (MNIST) handwritten digit database yielded a 92% recognition rate, thereby demonstrating the applicability of the proposed synaptic device to the neuromorphic system.

Today, there is already a significant injection of renewable energies at the medium-voltage level, which requires the use of reliable monitoring methods. In addition to tracking electrical parameters such as line current or bus voltage magnitudes, precise knowledge of the active and reactive power feed-in is becoming increasingly relevant in order to provide the necessary information for optimization strategies at higher voltage levels. For this reason, we have developed a method to monitor the active and reactive power for the medium-voltage level with very low measurement density, which is based on artificial neural networks (ANN). The actual training of ANN is accomplished with photovoltaics (PV) and wind feed-in time series based on real weather data to ensure realistic monitoring of the injection. The presented method is applied to a German medium-voltage grid to evaluate the estimation accuracy.

This paper presents the reconstruction of a 73-year time series of the aerosol optical depth (AOD) at 500 nm at the subtropical high-mountain Izaña Atmospheric Observatory (IZO) located in Tenerife (Canary Islands, Spain). For this purpose, we have combined AOD estimates from artificial neural networks (ANNs) from 1941 to 2001 and AOD measurements directly obtained with a Precision Filter Radiometer (PFR) between 2003 and 2013. The analysis is limited to summer months (July–August–September), when the largest aerosol load is observed at IZO (Saharan mineral dust particles). The ANN AOD time series has been comprehensively validated against coincident AOD measurements performed with a solar spectrometer Mark-I (1984–2009) and AERONET (AErosol RObotic NETwork) CIMEL photometers (2004–2009) at IZO, obtaining a rather good agreement on a daily basis: Pearson coefficient, R, of 0.97 between AERONET and ANN AOD, and 0.93 between Mark-I and ANN AOD estimates. In addition, we have analysed the long-term consistency between ANN AOD time series and long-term meteorological records identifying Saharan mineral dust events at IZO (synoptical observations and local wind records). Both analyses provide consistent results, with correlations > 85 %. Therefore, we can conclude that the reconstructed AOD time series captures well the AOD variations and dust-laden Saharan air mass outbreaks on short-term and long-term timescales and, thus, it is suitable to be used in climate analysis. The AERONET Cimel sun photometer at Izaña has been calibrated by AERONET-EUROPE Calibration Service, financed by the Aerosol Cloud and TRace gas InfraStructure (ACTRIS) European Research Infrastructure Action (FP7/2007-2013 no. 262254). Financial support from the Spanish Ministry of Economy and Competitiveness (MINECO) and from the “Fondo Europeo de Desarrollo Regional” (FEDER) for project CGL2012-33576 is gratefully acknowledged.

Bu tezin amacı metan buhar reformu ile ilgili bilgi çıkarımı yapmak ve literatürden elde edilmiş verileri temsilen modeller oluşturmaktır. Deneysel veriler literatürde yayınlanmış makalelerden toplanmıştır. Metan dönüşümü, MATLAB'de yazılan bilgisayar kodlarınca oluşturulmuş karar ağacı sınıflandırması ve yapay sinir ağları kullanılarak çeşitli katalizör hazırlama ve operasyonel değişkenlerine bağlı bir fonksiyon gibi modellenmiştir. Metan dönüşümü için karar ağacı analizleri tüm verilere, Ni, Ru ve Rh metali tabanlı katalizörlere, emdirme yöntemli ve dolgulu reaktörlü verilere ayrı ayrı uygulanmıştır. Tüm verilerin analizi %20.83 eğitici hata yüzdesi ve %22.91 test hata yüzdesi ile sonuçlanmıştır. Ni metal tabanlı veriler için %21.41 eğitici hata yüzdesi ve %24.52 test hata yüzdesi elde edilmiştir. Rh metal tabanlı veriler için eğitici hata yüzdesi ve test hata yüzdesi %6.68 ve %8.93 bulunmuştur. Ru metal tabanlı veriler için %8.03 eğitici hata yüzdesi ve %14.77 test hata yüzdesi hesaplanmıştır. Emdirme yöntemli verilerin eğitici hata yüzdesi ve test hata yüzdesi %11.47 ve %14.5'tir. Dolgulu reaktör verilerinin eğitici hata yüzdesi ve test hata yüzdesi %20.01 ve %21.78'dir. Sinir ağı analizi de gerçekleştirilmiştir ve en uygun sinir ağı topolojisi, eğitici analizde `trainlm` ve test analizinde `trainbr` fonksiyonunun kullanıldığı, 59-16-16-1 (59 giren nöron, ilk ve ikinci saklı katmanın her birinde 16 nöron ve 1 sonuç; her katman için aktivasyon fonksiyonu hiperbolik tanjant sigmoid fonksiyonu) olarak bulunmuştur. Eğitici sinir ağı modelinin R2 ve RMSE değerleri 0.97 ve 6.03 bulunurken test sinir ağı modelininki 0.93 ve 8.78 çıkmıştır. Girenlerin önem analizi sonucunda, giriş değişkenleri arasında metal türü metan dönüşümü üzerinde en önemli etkiyi göstermiştir. Katalizör hazırlama metodu ise en az önemli değişken olarak bulunmuştur. Son olarak, en uygun model, deneylerin sonuçlarını veritabanındaki diğer deneyler kullanılarak tahmin etmeye zorlanmıştır. RMSE ve R2 %69.6 ve %9.03 çıkmıştır. Makaleler için de yapıldığında bu oranlar sırasıyla %44.07 ve %18.64 olarak bulunmuştur. The aim of this thesis is to extract a general knowledge about the steam reforming of methane and to create models representing the data accumulated in the literature. The experimental data were collected from published articles in the literature. The conversion of methane was modeled as a function of various catalyst preparation and operational variables using decision tree classification and artificial neural networks, which were created by writing computer codes in MATLAB environment. Decision tree analyses for methane conversion were performed for the entire data, for Ni, Ru and Rh based catalysts, incipient to wetness impregnation method data and packed bed reactor data, separately. Analysis of total data resulted in 20.83% training error and 22.91% testing error. 21.41% training error and 24.52% testing error were obtained for Ni metal based data. 6.68% and 8.93% errors were found for training and testing of Rh metal based data. 8.03% training error and 14.77% testing error were calculated for Ru metal based data. Training error and testing error of incipient to wetness impregnation method data were 11.47% and 14.50%. For packed bed reactor data, training error and testing error were 20.01% and 21.78. The neural network analysis was also performed and the optimal neural network topology was found as 59-16-16-1 (59 input neurons, 16 neurons each in the first and second hidden layers and 1 output; with the activation function of hyperbolic tangent sigmoid function for all the layers), where `trainlm` and `trainbr` functions were used for training and testing respectively. R2 and RMSE values of training were found to be 0.97 and 6.03, whereas they were 0.93 and 8.78 for testing. Then, an input significance analysis was performed and it was found that base metal type within the input variables had the most significant effect on the methane conversion while catalyst preparation method was the least important parameter. Finally, the optimal neural network was forced to predict the results of experiments by using the data of the other experiments in the database. RMSE and R2 were 69.6% and 9.03% for experiments and 44.07% and 18.64% for articles, respectively. 86

The prediction of diseases caused by viral infections is a complex medical task where many real data that consists of different variables must be employed. As known, COVID-19 is the most dangerous disease worldwide; nowhere, an effective drug has been found yet. To limit its spread, it is essential to find a rational method that shows the spread of this virus by relying on many infected people’s data. A model consisting of three artificial neural networks’ (ANN) functions was developed to predict COVID-19 separation in Iraq based on real infection data supplied by the public health department at the Iraqi Ministry of Health. The performance efficiency of this model was evaluated, where its performance efficiency reached 81.6% when employed four statistical error criteria as mean absolute percentage error (MAPE), root mean square error (RMSE), coefficient of determination (R2), and Nash-Sutcliffe coefficient (NC). The severity of the virus’s spread across Iraq was assessed in a short term (in the next 6 months), where the results show that the spread severity will intensify in this short term by 17.1%, and the average death cases will increase by 8.3%. These results clarified by creating spatial distribution maps for virus spread are simulated by employing a Geographic Information System (GIS) environment to be used as a useful database for developing plans for combating viruses in Iraq.

A significant amount of attention has been given to the design and synthesis of co‐crystals by both industry and academia because of its potential to change a molecule's physicochemical properties. Yet, difficulties arise when searching for adequate combinations of molecules (or coformers) to form co‐crystals, hampering the efficient exploration of the target's solid‐state landscape. This paper reports on the application of a data‐driven co‐crystal prediction method based on two types of artificial neural network models and co‐crystal data present in the Cambridge Structural Database. The models accept pairs of coformers and predict whether a co‐crystal is likely to form. By combining the output of multiple models of both types, our approach shows to have excellent performance on the proposed co‐crystal training and validation sets, and has an estimated accuracy of 80 % for molecules for which previous co‐crystallization data is unavailable., An artificial neural network calculates whether two molecules are likely to form a co‐crystal. The approach shows excellent performance on the proposed co‐crystal training and validation sets, and has an estimated accuracy of 80 % for molecules for which previous co‐crystallization data is unavailable. The figure illustrates the prediction of the newly discovered drug–drug co‐crystal containing ketoprofen and carbamazepine.

This paper presents an overview on applications of artificial neural networks (ANNs) to robot control systems and outlines the contributions of ANNs. Adaptive neurocontrol architectures are introduced and compared to traditional adaptive control methods.

In this paper we will show that certain networks called ‘ inhibition nets ’ may be regarded as cognitive agents drawing nonmonotonic inferences. It will be proven that the system CL (introduced by KLM in [Artificial Intelligence 44 (1990) 186–189]) of nonmonotonic logic is both sound and complete with respect to the inferences drawn by finite hierarchical inhibition nets. The latter class of inhibition nets is shown to correspond to the class of finite, normal, hierarchical logic programs concerning dynamics, and also to the class of binary, layered, input-driven artificial neural networks.

2nd World Conference on Business, Economics and Management (BEM) -- APR 25-28, 2013 -- Antalya, TURKEY WOS: 000335570200182 Artificial neural networks (ANN) have been widely used in recent years to model non-linear time series since ANN approach is a responsive method and does not require some assumptions such as normality or linearity. An important problem with using ANN for time series forecasting is to determine the number of neurons in hidden layer. There have been some approaches in the literature to deal with the problem of determining the number of neurons in hidden layer. A new ANN model was suggested which is called multiplicative neuron model (MNM) in the literature. MNM has only one neuron in hidden layer. Therefore, the problem of determining the number of neurons in hidden layer is automatically solved when MNM is employed. Also, MNM can produce accurate forecasts for non-linear time series. ANN models utilized for non-linear time series have generally autoregressive structures since lagged variables of time series are generally inputs of these models. On the other hand, it is a well-known fact that better forecasts for real life time series can be obtained from models whose inputs are lagged variables of error. In this study, a new recurrent multiplicative neuron neural network model is firstly proposed. In the proposed method, lagged variables of error are included in the model. Also, the problem of determining the number of neurons in hidden layer is avoided when the proposed method is used. To train the proposed neural network model, particle swarm optimization algorithm was used. To evaluate the performance of the proposed model, it was applied to a real life time series. Then, results produced by the proposed method were compared to those obtained from other methods. It was observed that the proposed method has superior performance to existing methods. (C) 2014 The Authors. Published by Elsevier Ltd. Selection and peer review under responsibility of Organizing Committee of BEM 2013. SciencePark Res, Org & Counseling Ltd, Acad World Educ & Res Ctr, Hacettepe Univ, Eastern Mediterranean Univ, Near E Univ, Zirve Univ

Background Sunflower seeds are in the top five most abundant oilseeds in the world, as well as sunflower oil in the edible oils group. Recently, increasing attention has been paid to cold-pressed sunflower oil because less processing is involved and no solvent is used. The present study was carried out to investigate dimensions (length, width, thickness), firmness, general (moisture content and hull content, mass of 1000 seeds), gravimetric (true and bulk density, porosity) and geometric characteristics (equivalent diameter, surface area, seed volume, sphericity) of 20 new sunflower hybrid seeds. Steps to determine most of these parameters are quite simple and easy since the process does not require long time or special equipment. Results Principal component analysis and cluster analysis confirmed differences in the mentioned characteristics between oily and confectionary sunflower hybrid seeds. One of the major differences between two groups of samples was in extraction oil yield. Mechanical extraction oil yield of the oily hybrid seeds was significantly (P ˂ 0.05) higher (from 68.72 ± 4.21% to 75.61 ± 1.99%) compared to confectionary hybrids (from 20.10 ± 2.82% to 39.91 ± 6.23%). Extraction oil yield values are known only after oil extraction. Conclusion Knowledge of the extraction oil yield value before the mechanical extraction enables better management of the process. By application of the artificial neural network approach, an optimal neural network model was developed. The developed model showed a good generalization capability to predict the mechanical extraction oil yield of new sunflower hybrids based on the experimental data, which was a main goal of this paper. © 2021 Society of Chemical Industry.

Sewage sludge hydrochars (SSHs), which are produced by hydrothermal carbonization (HTC), offer a high calorific value to be applied as a biofuel. However, HTC is a complex processand the properties of the resulting product depend heavily on the process conditions and feedstock composition. In this work, we have applied artificial neural networks (ANNs) to contribute to the production of tailored SSHs for a specific application and with optimum properties. We collected data from the published literature covering the years 2014–2021, which was then fed into different ANN models where the input data (HTC temperature, process time, and the elemental content of hydrochars) were used to predict output parameters (higher heating value, (HHV) and solid yield (%)). The proposed ANN models were successful in accurately predicting both HHV and contents of C and H. While the model NN1 (based on C, H, O content) exhibited HHV predicting performance with R2 = 0.974, another model, NN2, was also able to predict HHV with R2 = 0.936 using only C and H as input. Moreover, the inverse model of NN3 (based on H, O content, and HHV) could predict C content with an R2 of 0.939.

Cod liver oil is a popular dietary supplement marketed as a rich source of omega-3 fatty acids as well as vitamins A and D. Due to its high market price, cod liver oil is vulnerable to adulteration with lower priced vegetable oils. In this study, 1H and 13C nuclear magnetic resonance spectroscopy, Fourier transform infrared spectroscopy, and gas chromatography (coupled to a flame ionization detector) were used in combination with multivariate statistics to determine cod liver oil adulteration with common vegetable oils (sunflower and canola oils). Artificial neural networks (ANN) were able to differentiate adulteration levels based on infrared spectra with a detection limit of 0.22% and a root mean square error of prediction (RMSEP) of 0.86%. ANN models using 1H NMR and 13C NMR data yielded detection limits of 3.0% and 1.8% and RMSEPs of 2.7% and 1.1%, respectively. In comparison, the ANN model based on fatty acid profiles determined by gas chromatography achieved a detection limit of 0.81% and an RMSEP of 1.1%. The approach of using spectroscopic techniques in combination with multivariate statistics can be regarded as a promising tool for the authentication of cod liver oil and may pave the way for a holistic quality assessment of fish oils. Graphical abstract Electronic supplementary material The online version of this article (10.1007/s00216-019-02063-y) contains supplementary material, which is available to authorized users.

Premature all-cause mortality is high in patients receiving peritoneal dialysis (PD). The accurate and early prediction of mortality is critical and difficult. Three prediction models, the logistic regression (LR) model, artificial neural network (ANN) classic model and a new structured ANN model (ANN mixed model), were constructed and evaluated using a receiver operating characteristic (ROC) curve analysis. The permutation feature importance was used to interpret the important features in the ANN models. Eight hundred fifty-nine patients were enrolled in the study. The LR model performed slightly better than the other two ANN models on the test dataset; however, in the total dataset, the ANN models fit much better. The ANN mixed model showed the best prediction performance, with area under the ROC curves (AUROCs) of 0.8 and 0.79 for the 6-month and 12-month datasets. Our study showed that age, diastolic blood pressure (DBP), and low-density lipoprotein cholesterol (LDL-c) levels were common risk factors for premature mortality in patients receiving PD. Our ANN mixed model had incomparable advantages in fitting the overall data characteristics, and age is a steady risk factor for premature mortality in patients undergoing PD. Otherwise, DBP and LDL-c levels should receive more attention for all-cause mortality during follow-up.

The objective of the present study was to evaluate the use of artificial neural networks (ANNs) in the development of a new chemometric model that will be able to simultaneously distinguish and quantify the percentage of the crystalline and the neat amorphous drug located within the drug-rich amorphous zones formed in an amorphous solid dispersion (ASD) system. Attenuated total reflectance Fourier-transform infrared (ATR-FTIR) spectroscopy was used, while Rivaroxaban (RIV, drug) and Soluplus® (SOL, matrix-carrier) were selected for the preparation of a suitable ASD model system. Adequate calibration and test sets were prepared by spiking different percentages of the crystalline and the amorphous drug in the ASDs (prepared by the melting - quench cooling approach), while a 24 full factorial experimental design was employed for the screening of ANN's structure and training parameters as well as spectra region selection and data preprocessing. Results showed increased prediction performance, measured based on the root mean squared error of prediction (RMSEp) for the test sample, for both the crystalline (RMSEp (crystal) = 0.86) and the amorphous (RMSEp (amorphous) = 2.14) drug. Comparison with traditional regression techniques, such as partial least square and principle component regressions, revealed the superiority of ANNs, indicating that in cases of high structural similarity between the investigated compounds (i.e., the crystalline and the amorphous forms of the same compound) the implementation of more powerful/sophisticated regression techniques, such as ANNs, is mandatory., Graphical abstract Unlabelled Image

Significance: Diffuse reflectance spectroscopy (DRS) is frequently used to assess oxygen saturation and hemoglobin concentration in living tissue. Methods solving the inverse problem may include time-consuming nonlinear optimization or artificial neural networks (ANN) determining the absorption coefficient one wavelength at a time. Aim: To present an ANN-based method that directly outputs the oxygen saturation and the hemoglobin concentration using the shape of the measured spectra as input. Approach: A probe-based DRS setup with dual source-detector separations in the visible wavelength range was used. ANNs were trained on spectra generated from a three-layer tissue model with oxygen saturation and hemoglobin concentration as target. Results: Modeled evaluation data with realistic measurement noise showed an absolute root-mean-square (RMS) deviation of 5.1% units for oxygen saturation estimation. The relative RMS deviation for hemoglobin concentration was 13%. This accuracy is at least twice as good as our previous nonlinear optimization method. On blood-intralipid phantoms, the RMS deviation from the oxygen saturation derived from partial oxygen pressure measurements was 5.3% and 1.6% in two separate measurement series. Results during brachial occlusion showed expected patterns. Conclusions: The presented method, directly assessing oxygen saturation and hemoglobin concentration, is fast, accurate, and robust to noise. (C) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Funding Agencies|Swedish Research CouncilSwedish Research Council [2014-6141]; Swedens innovation agency VINNOVA via the program MedTech4Health [2016-02211]; Swedens innovation agency VINNOVA via the program Swelife and MedTech4Health [2017-01435]

Introduction A machine learning technique that imitates neural system and brain can provide better than traditional methods like logistic regression for survival prediction and create an algorithm by determining influential factors. Aim To determine the influential factors on survival time of palliative care cancer patients and to compare two statistical methods for better prediction of survival. Methods One-year data is gathered from the patients that we followed in the palliative care clinic of our hospital (2017-2018) (n = 189). All data were retrospectively evaluated. After descriptive statistics, we used Pearson and Spearman correlations for parametric and non-parametric variables. The Artificial Neural Networks (ANN) and logistic regression model were applied to parameters which have a significant correlation with short survival. Results Significantly correlated variables with short survival were Palliative Performance Scale (PPS), Edmonton Symptom Assessment System (ESAS), Karnofsky Performance Scale (KPS), brain, liver, and distant metastasis, hemogram parameters, cero-reactive protein (CRP) and albumin (ALB). ANN model showed 89.3% prediction accuracy while the logistic regression model showed 73.0%. ANN model achieved a better AUC value of 0.86 than logistic regression model (0.76). Discussion There are several prognostic evaluation tools such as PPS, KPS, CRP, albumin, leukocytes, neutrophil were reported several studies as survival-related parameters in logistic regression models, also. Many studies compare ANN with logistic regression. When we evaluated these parameters totally, we observed the same relations with survival then we used the same parameters in the ANN model. The effectivity of the survival prediction models can be improved with the use of ANN. Conclusion ANN provides a more accurate estimation than logistic regression. ANN model is an important statistical method for survival prediction of cancer patients.

Industry 4.0 is all about interconnectivity, sensor-enhanced process control, and data-driven systems. Process analytical technology (PAT) such as online nuclear magnetic resonance (NMR) spectroscopy is gaining in importance, as it increasingly contributes to automation and digitalization in production. In many cases up to now, however, a classical evaluation of process data and their transformation into knowledge is not possible or not economical due to the insufficiently large datasets available. When developing an automated method applicable in process control, sometimes only the basic data of a limited number of batch tests from typical product and process development campaigns are available. However, these datasets are not large enough for training machine-supported procedures. In this work, to overcome this limitation, a new procedure was developed, which allows physically motivated multiplication of the available reference data in order to obtain a sufficiently large dataset for training machine learning algorithms. The underlying example chemical synthesis was measured and analyzed with both application-relevant low-field NMR and high-field NMR spectroscopy as reference method. Artificial neural networks (ANNs) have the potential to infer valuable process information already from relatively limited input data. However, in order to predict the concentration at complex conditions (many reactants and wide concentration ranges), larger ANNs and, therefore, a larger training dataset are required. We demonstrate that a moderately complex problem with four reactants can be addressed using ANNs in combination with the presented PAT method (low-field NMR) and with the proposed approach to generate meaningful training data. Graphical abstract Electronic supplementary material The online version of this article (10.1007/s00216-020-02687-5) contains supplementary material, which is available to authorized users.

MicroRNAs (miRNAs) are short non-coding RNAs that regulate expression of target messenger RNAs (mRNAs) post-transcriptionally. Understanding the precise regulatory role of miRNAs is of great interest since miRNAs have been shown to play an important role in development, diseases, and other biological processes. Early work on miRNA target prediction has focused on static sequence-driven miRNA-mRNA complementarity. However, recent research also utilizes expression-level data to study context-dependent regulation effects in a more dynamic, physiologically-relevant setting. Methods: We propose a novel artificial neural network (ANN) based method, named Tiresias, to predict such targets in a context-dependent manner by combining sequence and expression data. In order to predict the interacting pairs among miRNAs and mRNAs and their regulatory weights, we develop a two-stage ANN and present how to train it appropriately. Tiresias is designed to study various regulation models, ranging from a simple linear model to a complex non-linear model. Tiresias has a single hyper-parameter to control the sparsity of miRNA-mRNA interactions, which we optimize using Bayesian optimization. Results: Tiresias performs better than existing computational methods such as GenMiR++, Elastic Net, and PIMiM, achieving an F1 score of >0.8 for a certain level of regulation strength. For the TCGA breast invasive carcinoma dataset, Tiresias results in the rate of up to 82% in detecting the experimentally-validated interactions between miRNAs and mRNAs, even if we assume that true regulations may result in a low level of regulation strength. Conclusion: Tiresias is a two-stage ANN, computational method that deciphers context-dependent microRNA regulatory interactions. Experiment results demonstrate that Tiresias outperforms existing solutions and can achieve a high F1 score. Source code of Tiresias is available at https://bitbucket.org/cellsandmachines/.

Günümüzde endüstrinin her alanında geleceğe ilişkin tahmin ve analizler tüm dünya ülkelerinde gerek özel, gerekse bilim dünyasında kabul görmüş ve bu konularda araştırmalar yapılmış ve yapılmaktadır. Herhangi bir değişkenin gelecekte alacağı değer, o değişkenin geçmiş dönemlerdeki gözlem değerleri ile oluşturulan modeller yardımıyla tahmin edilmektedir. Yapay sinir ağları (YSA) son dönemlerde diğer bilimsel alanlarda olduğu gibi tahmin alanında da sıkça kullanılan bir yöntem haline gelmiştir. Bu çalışmada Türkiye kağıt-karton sanayisinin ihracat rakamları yapay sinir ağları kullanılarak tahmin edilmiştir. YSA girdi değişkeni olarak Türkiye kağıt-karton üretimi, atık kağıt, endüstriyel odun, tomruk üretimi, nüfus, GSYİH, TÜFE, ÜFE, döviz kurları ve ekonomik büyüme rakamları kullanılmıştır. Sonuçlar kağıt-karton ihracatının 2025 yılında 2011 yılına oranla %48,2’lik bir artış, 2015 yılına göre ise %3’lük bir azalış göstereceğini öngörmektedir.

The paper presents simulation of the cutting force components for machining of magnesium alloy AZ91HP. The simulation employs the Black Box model. The closest match to (input and output) data obtained from the machining process was determined. The simulation was performed with the use of the Statistica programme with the application of neural networks: RBF (Radial Basis Function) and MLP (Multi-Layered Perceptron).

Automatic sleep stage scoring based on deep neural networks has come into focus of sleep researchers and physicians, as a reliable method able to objectively classify sleep stages would save human resources and simplify clinical routines. Due to novel open-source software libraries for machine learning, in combination with enormous recent progress in hardware development, a paradigm shift in the field of sleep research towards automatic diagnostics might be imminent. We argue that modern machine learning techniques are not just a tool to perform automatic sleep stage classification, but are also a creative approach to find hidden properties of sleep physiology. We have already developed and established algorithms to visualize and cluster EEG data, facilitating first assessments on sleep health in terms of sleep-apnea and consequently reduced daytime vigilance. In the following study, we further analyze cortical activity during sleep by determining the probabilities of momentary sleep stages, represented as hypnodensity graphs and then computing vectorial cross-correlations of different EEG channels. We can show that this measure serves to estimate the period length of sleep cycles and thus can help to find disturbances due to pathological conditions., Highlights • Artificial intelligence leads to a paradigm shift in sleep research. • Sleep stage scoring with deep neural networks is robust and reliable. • Neural networks find embeddings of EEG raw data. • Probabilities of momentary sleep stages can be visualized as hypnodensity graphs. • Cross-correlations of sleep stage probability vectors provide an estimate of sleep cycle period length.

Highlights • Blood DNA methylation profiles of 1156 individuals were assessed for age correlation. • Stepwise regression identified 23 age-associated CpG sites in DNA from blood. • A machine learning model based on 16 markers predicted age with a mean error of 3.8 years. • The model predicted age successfully for twins and ‘diseased’ individuals. • A new NGS-based method was combined with machine learning for age prediction., The ability to estimate the age of the donor from recovered biological material at a crime scene can be of substantial value in forensic investigations. Aging can be complex and is associated with various molecular modifications in cells that accumulate over a person’s lifetime including epigenetic patterns. The aim of this study was to use age-specific DNA methylation patterns to generate an accurate model for the prediction of chronological age using data from whole blood. In total, 45 age-associated CpG sites were selected based on their reported age coefficients in a previous extensive study and investigated using publicly available methylation data obtained from 1156 whole blood samples (aged 2–90 years) analysed with Illumina’s genome-wide methylation platforms (27 K/450 K). Applying stepwise regression for variable selection, 23 of these CpG sites were identified that could significantly contribute to age prediction modelling and multiple regression analysis carried out with these markers provided an accurate prediction of age (R2 = 0.92, mean absolute error (MAE) = 4.6 years). However, applying machine learning, and more specifically a generalised regression neural network model, the age prediction significantly improved (R2 = 0.96) with a MAE = 3.3 years for the training set and 4.4 years for a blind test set of 231 cases. The machine learning approach used 16 CpG sites, located in 16 different genomic regions, with the top 3 predictors of age belonged to the genes NHLRC1, SCGN and CSNK1D. The proposed model was further tested using independent cohorts of 53 monozygotic twins (MAE = 7.1 years) and a cohort of 1011 disease state individuals (MAE = 7.2 years). Furthermore, we highlighted the age markers’ potential applicability in samples other than blood by predicting age with similar accuracy in 265 saliva samples (R2 = 0.96) with a MAE = 3.2 years (training set) and 4.0 years (blind test). In an attempt to create a sensitive and accurate age prediction test, a next generation sequencing (NGS)-based method able to quantify the methylation status of the selected 16 CpG sites was developed using the Illumina MiSeq® platform. The method was validated using DNA standards of known methylation levels and the age prediction accuracy has been initially assessed in a set of 46 whole blood samples. Although the resulted prediction accuracy using the NGS data was lower compared to the original model (MAE = 7.5 years), it is expected that future optimization of our strategy to account for technical variation as well as increasing the sample size will improve both the prediction accuracy and reproducibility.

In recent years, state‐of‐the‐art computational modeling of physical and chemical systems has shown itself to be an invaluable resource in the prediction of the properties and behavior of functional materials. However, construction of a useful computational model for novel systems in both academic and industrial contexts often requires a great depth of physicochemical theory and/or a wealth of empirical data, and a shortage in the availability of either frustrates the modeling process. In this work, computational intelligence is instead used, including artificial neural networks and evolutionary computation, to enhance our understanding of nature‐inspired superhydrophobic behavior. The relationships between experimental parameters (water droplet volume, weight percentage of nanoparticles used in the synthesis of the polymer composite, and distance separating the superhydrophobic surface and the pendant water droplet in adhesive force measurements) and multiple objectives (water droplet contact angle, sliding angle, and adhesive force) are built and weighted. The obtained optimal parameters are consistent with the experimental observations. This new approach to materials modeling has great potential to be applied more generally to aid design, fabrication, and optimization for myriad functional materials.

Barley straw is a lignocellulosic biomass that can be used to obtain value-added products for industrial applications. Barley straw hydrolysis with sodium sulfite facilitates the production of lignosulfonates. In this work, the delignification process of barley straw by solubilizing lignin through sulfite method was studied. Response surface methodology and artificial neural network were used to develop predictive models for simulation and optimization of delignification process of barley straw. The influence of parameters over sulfite concentration (1.0 to 10.0%), particle size (8 to 20), and reaction time (30 to 90 min) on total percentage of solubilized material was investigated through a three level three factor (33) full factorial central composite design with the help of Matlab® ver. 8.1. The results show that particle size and sulfite concentration have the most significant effect on delignification process. Both techniques, response surface methodology and artificial neural networks, predicted the lignosulfonate yield adequately, although the artificial neural network technique produced a better fit (R2 = 0.9825) against the response surface methodology (R2 = 0.9290). Based on these findings, this study can be used as a guide to forecast the potential production of lignosulfonates from barley straw using different experimental conditions.

In this paper two procedures are developed for the identification of the parameters contained in an orthotropic elastic-plastic-hardening model for free standing foils, particularly of paper and paperboard. The experimental data considered are provided by cruciform tests and digital image correlation. A simplified version of the constitutive model proposed by Xia et al. (Int J Solids Struct 39:4053---4071, 2002) is adopted. The inverse analysis is comparatively performed by the following alternative computational methodologies: (a) mathematical programming by a trust-region algorithm; (b) proper orthogonal decomposition and artificial neural network. The second procedure rests on preparatory once-for-all computations and turns out to be applicable economically and routinely in industrial environments.

In relation to the situation caused by the pandemic, which may also take place in the future, there is a need to find effective solutions to improve the economic situation of the floristry industry. The production and sale of flowers is time-consuming and long-term. Therefore, any information that causes the impossibility of selling the plants will result in a reduction of profitability or bankruptcy of such companies. Research on rationally utilizing biowaste from plant cultivation as well as unsold flowers for environmental protection and effective use of their potential as a raw material for bioenergy production were examined in this article. The aim of this study was to analyze the energetic potential of the biodegradable fraction of waste from floriculture. The trials included floricultural waste containing the stems, leaves and flowers of different species and hybrid tulips (Tulipa L.), roses (Rosa L.), sunflowers (Helianthus L.) and chrysanthemums (Dendranthema Des Moul.). Their biogas and methane production as well as heat of combustion were determined experimentally. The calorific value was calculated on the basis of results from selected floricultural waste and its chemical composition. The biogas production was tested on different levels of plant material fragmentation (chaff, macerate) in fermentation processes with two ranges of temperature (meso- and thermophilic fermentation). The presented results show that the highest calorific values were determined for dry stems of roses (18,520 kJ/kg) and sunflowers (18,030 kJ/kg). In turn, the lowest were obtained for dried chrysanthemums and tulips, for which the heating value reached 15,560 kJ/kg and 15,210 kJ/kg. In addition, based on one ton of the fresh mass of biowaste from floriculture, the largest biogas production including the control was obtained from the chrysanthemum chaff by mesophilic anaerobic digestion. Moreover, the largest volume of methane was received by thermophilic anaerobic digestion of roses. The highest content of biomethane (56.68%) was reached by thermophilic fermentation of roses. The energy production of the analyzed substrates was also calculated, based on the amount of biogas produced in the containers for anaerobic digestion. Additionally, a deep neural network model, which predicted the production of methane gas, was created. Owing to the properties of the network, the level of significance of variables used for modelling and prediction of biogas production was determined. The neural modelling process was carried out with the use of the H2O program.

Laser speckle contrast imaging (LSCI) enables video rate imaging of blood flow. However, its relation to tissue blood perfusion is nonlinear and depends strongly on exposure time. By contrast, the perfusion estimate from the slower laser Doppler flowmetry (LDF) technique has a relationship to blood perfusion that is better understood. Multiexposure LSCI (MELSCI) enables a perfusion estimate closer to the actual perfusion than that using a single exposure time. We present and evaluate a method that utilizes contrasts from seven exposure times between 1 and 64 ms to calculate a perfusion estimate that resembles the perfusion estimate from LDF. The method is based on artificial neural networks (ANN) for fast and accurate processing of MELSCI contrasts to perfusion. The networks are trained using modeling of Doppler histograms and speckle contrasts from tissue models. The importance of accounting for noise is demonstrated. Results show that by using ANN, MELSCI data can be processed to LDF perfusion with high accuracy, with a correlation coefficient R = 1.000 for noise-free data, R = 0.993 when a moderate degree of noise is present, and R = 0.995 for in vivo data from an occlusion-release experiment.

Background Each year, approximately 800,000 people die by suicide worldwide, accounting for 1–2 in every 100 deaths. It is always a tragic event with a huge impact on family, friends, the community and health professionals. Unfortunately, suicide prevention and the development of risk assessment tools have been hindered by the complexity of the underlying mechanisms and the dynamic nature of a person’s motivation and intent. Many of those who die by suicide had contact with health services in the preceding year but identifying those most at risk remains a challenge. Objective To explore the feasibility of using artificial neural networks with routinely collected electronic health records to support the identification of those at high risk of suicide when in contact with health services. Methods Using the Secure Anonymised Information Linkage Databank UK, we extracted the data of those who died by suicide between 2001 and 2015 and paired controls. Looking at primary (general practice) and secondary (hospital admissions) electronic health records, we built a binary feature vector coding the presence of risk factors at different times prior to death. Risk factors included: general practice contact and hospital admission; diagnosis of mental health issues; injury and poisoning; substance misuse; maltreatment; sleep disorders; and the prescription of opiates and psychotropics. Basic artificial neural networks were trained to differentiate between the suicide cases and paired controls. We interpreted the output score as the estimated suicide risk. System performance was assessed with 10x10-fold repeated cross-validation, and its behavior was studied by representing the distribution of estimated risk across the cases and controls, and the distribution of factors across estimated risks. Results We extracted a total of 2604 suicide cases and 20 paired controls per case. Our best system attained a mean error rate of 26.78% (SD 1.46; 64.57% of sensitivity and 81.86% of specificity). While the distribution of controls was concentrated around estimated risks < 0.5, cases were almost uniformly distributed between 0 and 1. Prescription of psychotropics, depression and anxiety, and self-harm increased the estimated risk by ~0.4. At least 95% of those presenting these factors were identified as suicide cases. Conclusions Despite the simplicity of the implemented system, the proposed methodology obtained an accuracy like other published methods based on specialized questionnaire generated data. Most of the errors came from the heterogeneity of patterns shown by suicide cases, some of which were identical to those of the paired controls. Prescription of psychotropics, depression and anxiety, and self-harm were strongly linked with higher estimated risk scores, followed by hospital admission and long-term drug and alcohol misuse. Other risk factors like sleep disorders and maltreatment had more complex effects.

The genus Leptosphaeria contains numerous fungi that cause the symptoms of asthma and also parasitize wild and crop plants. In search of a robust and universal forecast model, the ascospore concentration in air was measured and weather data recorded from 1 March to 31 October between 2006 and 2012. The experiment was conducted in three European countries of the temperate climate, i.e., Ukraine, Poland, and the UK. Out of over 150 forecast models produced using artificial neural networks (ANNs) and multivariate regression trees (MRTs), we selected the best model for each site, as well as for joint two-site combinations. The performance of all computed models was tested against records from 1 year which had not been used for model construction. The statistical analysis of the fungal spore data was supported by a comprehensive study of both climate and land cover within a 30-km radius from the air sampler location. High-performance forecasting models were obtained for individual sites, showing that the local micro-climate plays a decisive role in biology of the fungi. Based on the previous epidemiological studies, we hypothesized that dew point temperature (DPT) would be a critical factor in the models. The impact of DPT was confirmed only by one of the final best neural models, but the MRT analyses, similarly to the Spearman’s rank test, indicated the importance of DPT in all but one of the studied cases and in half of them ranked it as a fundamental factor. This work applies artificial neural modeling to predict the Leptosphaeria airborne spore concentration in urban areas for the first time. Electronic supplementary material The online version of this article (10.1007/s00484-018-1500-z) contains supplementary material, which is available to authorized users.

The longevity and robustness of bioreactors used for wastewater treatment is determined by the activity of the microorganisms under steady and transient loading conditions. Two identical continuously operated inverse fluidized bed bioreactors (IFB), IFB R1 and IFB R2, were tested for sulphate removal under the same operating conditions for 140 d (Periods I–IV). Later, IFB R1 was used as the control reactor (Period V), while IFB R2 was operated under feast (Period V-A) and famine (Period V-B) feeding conditions for 66 d. The sulphate removal efficiency was comparable in both IFB

This paper describes artificial neural network ( ANN) based prediction of the response of a fiber optic sensor using evanescent field absorption (EFA). The sensing probe of the sensor is made up a bundle of five PCS fibers to maximize the interaction of evanescent field with the absorbing medium. Different backpropagation algorithms are used to train the multilayer perceptron ANN. The Levenberg-Marquardt algorithm, as well as the other algorithms used in this work successfully predicts the sensor responses.

This review paper discusses the application of hybrid deep learning techniques for predicting the charging state of electric vehicles. The paper highlights the importance of accurate predictions for the efficient management of electric vehicle charging stations. The review focuses on the use of recursive neural networks (RNNs) and the gated recurrent unit (GRU) framework in hybrid deep learning models, which have shown promising results in previous studies. In addition to hybrid deep learning, the paper also examines the use of support vector machines (SVMs) and artificial neural networks (ANNs) in charging state prediction. The strengths and weaknesses of these different approaches are analyzed and compared. The paper concludes that hybrid deep learning models, particularly those using RNNs and GRUs, are a promising approach for accurately predicting electric vehicle charging states. The paper also suggests potential areas for future research to further improve the accuracy and efficiency of charging state predictions.

An experimental mixture design coupled with data analysis by means of both response surface methodology (RSM) and artificial neural networks (ANNs) followed by multiple response optimization through a desirability function, was applied to the production of δ-endotoxins from Bacillus thuringiensis var. kurstaki. The composition of a culture medium was defined by testing three regional effluents: milky effluent, beer wastewater and sugar cane molasses. Both RSM and ANNs accomplished the goal pursued in this work, by predicting the optimal mixture of the effluents. ANNs provided more reliable results due to the complexity of the models to be fitted. The optimal selected blend was: 74%, 26% and 0%, respectively for each the above-mentioned effluents. Fil: Moreira, Guilherme. Universidad Nacional del Litoral; Argentina Fil: Micheloud, Gabriela Analia. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - Santa Fe. Instituto de Agrobiotecnología del Litoral. Universidad Nacional del Litoral. Instituto de Agrobiotecnología del Litoral; Argentina Fil: Beccaria, Alejandro José. Universidad Nacional del Litoral; Argentina Fil: Goicoechea, Hector Casimiro. Universidad Nacional del Litoral; Argentina

U ovom radu opisan je razvoj računalnog agenta za igru zmija, čija pravila su objašnjena na početku rada. Agent je razvijen uz pomoć neuronskih mreža i genetskih algoritama, a arhitektura mreže i način odvijanja selekcije, križanja i mutacije opisan je u odgovarajućim poglavljima. Izrađeno je grafičko korisničko sučelje koje omogućuje igranje od strane čovjeka ili praćenje procesa treniranja agenta. Grafički su prikazani dobiveni rezultati u procesu treniranja i testiranja. This paper describes the development of a computer agent for the snake game, the rules of which are explained at the beginning of the paper. The agent is developed from neural networks and genetic algorithms, the architecture of the network and the methods of selection, crossover and mutation are described in the corresponding chapters. A graphical user interface has been created that allows playing by a human or monitoring the process of training an agent. The results obtained in the process of training and testing are shown graphically.

Pirmoje darbo dalyje analizuojami mokslinėje literatūroje taikyti metodai sukčiavimo rizikai vertinti, išskiriami reikšmingiausi nustatyti sukčiavimo riziką lemiantys veiksniai. Antroje darbo dalyje pristatomi surinkti viešai prieinami duomenys apie UAB „Bendras finansavimas“ antrinėje rinkoje parduodamas paskolas. Apibrėžiami tyrime naudoti duomenų imties tvarkymo bei sukčiavimo rizikos vertinimo metodai: atsitiktiniai miškai, atraminių vektorių mašinos, dirbtiniai neuroniniai tinklai. Trečioje darbo dalyje duomenys apdoroti pagrindinių komponenčių analizės būdu. Sudaryti atsitiktinių miškų, atraminių vektorių mašinų bei dirbtinių neuroninių tinklų modeliai naudojant tris skirtingas duomenų imtis: pradinį duomenų rinkinį (neapdorotą pagrindinių komponenčių analizės būdu); imtį su išskirtomis aštuoniomis pagrindinėmis komponentėmis; imtį su išskirtomis aštuoniomis pagrindinėmis komponentėmis bei kokybiniais tyrimo kintamaisiais. Atlikus tyrimą nustatyta, kad atsitiktinių miškų modelis su aštuonių pagrindinių komponenčių bei kokybinių kintamųjų duomenų failu tiksliausiai suklasifikuoja tyrimo objektą – sukčiavimo faktą. Reikšmingiausiais sukčiavimą lemiančiais veiksniais išskirti: bloga kredito istorija, įsipareigojimų ir pajamų santykis, paskolos suma, kliento nemokumo rizikos įvertis bei amžius., In the first part of the paper, the methods used in the scientific literature to assess the risk of fraud are analyzed and the main factors determining the risk of fraud are elaborated. The second part of the paper presents the collected publicly available data on the loans sold by UAB "Bendras finansavimas" on the secondary market. The data sample management and fraud risk assessment methods used in the study are defined: Random Forests, Support Vector Machines, Artificial Neural Networks. In the third part of the paper, the data are processed using principal component analysis. Models of Random Forests, Support Vector Machines and Artificial Neural Networks were created with three different data samples: Primary dataset (without implemented principal component analysis); a sample with eight extracted principal components; sample with eight principal components and qualitative research variables. After conducting the research, it was found that the Random Forest model with the dataset of eight principal components and qualitative variables most accurately classified the object of study - the fact of fraud. The main factors determining fraud are: bad credit history, debt service-to-income ratio, the amount of the loan, the client's assessment of the risk of fraud and age.

Today, the chemical corrosion of metals is one of the main problems of large productions, especially in the oil and gas industries. Due to massive downtime connected to corrosion failures, pipeline corrosion is a central issue in many oil and gas industries. Therefore, the determination of the corrosion progress of oil and gas pipelines is crucial for monitoring the reliability and alleviation of failures that can positively impact health, safety, and the environment. Gas transmission and distribution pipes and other structures buried (or immersed) in an electrolyte, by the existing conditions and due to the metallurgical structure, are corroded. After some time, this disrupts an active system and process by causing damage. The worst corrosion for metals implanted in the soil is in areas where electrical currents are lost. Therefore, cathodic protection (CP) is the most effective method to prevent the corrosion of structures buried in the soil. Our aim in this paper is first to investigate the effect of stray currents on failure rate using the condition index, and then to estimate the remaining useful life of CP gas pipelines using an artificial neural network (ANN). Predicting future values using previous data based on the time series feature is also possible. Therefore, this paper first uses the general equipment condition monitoring method to detect failures. The time series model of data is then measured and operated by neural networks. Finally, the amount of failure over time is determined.

Cholesteryl esters have antimicrobial activity and likely contribute to the innate immunity system. Improved separation techniques are needed to characterize these compounds. In this study, optimization of the reversed-phase high-performance liquid chromatography separation of six analyte standards (four cholesteryl esters plus cholesterol and tri-palmitin) was accomplished by modeling with an artificial neural network–genetic algorithm (ANN-GA) approach. A fractional factorial design was employed to examine the significance of four experimental factors: organic component in the mobile phase (ethanol and methanol), column temperature, and flow rate. Three separation parameters were then merged into geometric means using Derringer’s desirability function and used as input sources for model training and testing. The use of genetic operators proved valuable for the determination of an effective neural network structure. Implementation of the optimized method resulted in complete separation of all six analytes, including the resolution of two previously co-eluting peaks. Model validation was performed with experimental responses in good agreement with model-predicted responses. Improved separation was also realized in a complex biological fluid, human milk. Thus, the first known use of ANN-GA modeling for improving the chromatographic separation of cholesteryl esters in biological fluids is presented and will likely prove valuable for future investigators involved in studying complex biological samples. Figure ANN-derived response surface plot for two interacting factors and overall response Electronic supplementary material The online version of this article (doi:10.1007/s00216-010-3778-5) contains supplementary material, which is available to authorized users.

The use of data mining and machine learning tools is becoming increasingly common. Their usefulness is mainly noticeable in the case of large datasets, when information to be found or new relationships are extracted from information noise. The development of these tools means that datasets with much fewer records are being explored, usually associated with specific phenomena. This specificity most often causes the impossibility of increasing the number of cases, and that can facilitate the search for dependences in the phenomena under study. The paper discusses the features of applying the selected tools to a small set of data. Attempts have been made to present methods of data preparation, methods for calculating the performance of tools, taking into account the specifics of databases with a small number of records. The techniques selected by the author are proposed, which helped to break the deadlock in calculations, i.e., to get results much worse than expected. The need to apply methods to improve the accuracy of forecasts and the accuracy of classification was caused by a small amount of analysed data. This paper is not a review of popular methods of machine learning and data mining; nevertheless, the collected and presented material will help the reader to shorten the path to obtaining satisfactory results when using the described computational methods

This paper presents a novel Transformer-based facial landmark localization network named Localization Transformer (LOTR). The proposed framework is a direct coordinate regression approach leveraging a Transformer network to better utilize the spatial information in the feature map. An LOTR model consists of three main modules: 1) a visual backbone that converts an input image into a feature map, 2) a Transformer module that improves the feature representation from the visual backbone, and 3) a landmark prediction head that directly predicts the landmark coordinates from the Transformer's representation. Given cropped-and-aligned face images, the proposed LOTR can be trained end-to-end without requiring any post-processing steps. This paper also introduces the smooth-Wing loss function, which addresses the gradient discontinuity of the Wing loss, leading to better convergence than standard loss functions such as L1, L2, and Wing loss. Experimental results on the JD landmark dataset provided by the First Grand Challenge of 106-Point Facial Landmark Localization indicate the superiority of LOTR over the existing methods on the leaderboard and two recent heatmap-based approaches. On the WFLW dataset, the proposed LOTR framework demonstrates promising results compared with several state-of-the-art methods. Additionally, we report the improvement in state-of-the-art face recognition performance when using our proposed LOTRs for face alignment., Accepted for publication in IEEE Access

Financial market forecasting is a challenging and complex task due to the sensitivity of the market to various factors such as political, economic, and social factors. However, recent advances in machine learning and computation technology have led to an increased interest in using deep learning for forecasting financial data. One the one hand, the famous efficient market hypothesis states that the market is so efficient that no one can consistently benefit from it, and the random walk theory suggests that asset prices are unpredictable based on historical data. On the other hand, previous research has shown that financial time series can be forecasted to some extent using artificial neural networks (ANNs). Despite being a relatively new addition to financial research with less study than the traditional models such as moving averages and linear regression models, ANNs have been shown to outperform the traditional models to some extent. Hence, considering the efficient market hypothesis and the random walk theory, there is a knowledge gap on whether neural networks can be used for financial time series prediction. This paper explores the use of ANNs, specifically recurrent neural networks, to predict financial time series data using a long short-term memory (LSTM) network model. The study will employ an experimental research strategy to construct and test an LSTM model to predict financial time series data, with the aim of examining its performance and evaluating it relative to other models and methods. For evaluating its performance, evaluation metrics are computed and the model is compared with a constructed simple moving average (SMA) model as well as other models in existing studies. The paper also explores the application and processing of transformed financial data, where it was found that achieving stationarity by data transformation was not necessary for the LSTM model to perform better. The study also found that the LSTM model outperformed the SMA model when hyperparameters were set to capture long-term dependencies. However, in the short-term, the SMA model outperformed the LSTM model.

Motivated by the idea that there should be a close relationship between biological significance and low power driving of spike neural networks (SNNs), this paper aims to focus on spike-frequency adaptation, which deviates significantly from existing biological meaningfulness, and develop a new spike-frequency adaptation with more biological characteristics. As a result, this paper proposes the sensoryadaptation method that reflects the mechanisms of the human sensory organs, and studies network architectures and neuron models for the proposed method. Next, this paper introduces a dedicated SNN simulator that can selectively apply the conventional spike-frequency adaptation and the proposed method, and provides the results of functional verification and effectiveness evaluation of the proposed method. Through intensive simulation, this paper reveals that the proposed method can produce a level of training and testing performance similar to the conventional method while significantly reducing the number of spikes to 32.66% and 45.63%, respectively. Furthermore, this paper contributes to SNN research by showing an example based on in-depth analysis that embedding biological meaning in SNNs may be closely related to the low-power driving characteristics of SNNs.

Recent years have seen an increase in events of drone incursion into airport terminal areas, leading to safety concerns and disruptions to airline operations. It is of great importance to identify the potential conflict, especially for those non-cooperative drones, as their intentions are always unknown. For the safe operation of air traffic, this paper proposes a conflict risk assessment method between non-cooperative drones and manned aircraft in the terminal area. First, the trajectory data of manned aircraft and drones are obtained. Real-time cylindrical protection zones are established around manned aircraft according to the separation interval for safe operation between the drone and the manned aircraft at different altitudes. Secondly, trajectory predictions for the manned aircraft and the drone are conducted, respectively. A quartile regression bidirectional gate recurrent unit neural network is proposed in this research for the trajectory prediction of the drones. The model integrates the bidirectional gated recurrent unit structure and the quartile regression structure. The performance indicators confirm the superiority of the proposed model. Based on the trajectory prediction results, it is then determined whether there is a conflict risk between the drone and manned aircraft by comparing the position distribution of the drone as well as the real-time cylindrical protection zone of the manned aircraft. The conflict probability between the drone and the manned aircraft is then calculated. The prediction accuracy of conflict probability is estimated by Monte Carlo simulation methods. The collision probability prediction accuracy of manned aircraft and drones at different flight stages and altitudes ranges from 73% to 97%, which shows the reliability of the proposed method. Finally, the collision probability between the drone and the manned aircraft at the closest encountering point and the estimated time to reach the closest encountering point are calculated. This paper predicts the conflict risk between the drone and manned aircraft, thus providing theoretical support for the safe operation of air transport in low-altitude environments.

Kroz ovaj rad je prikazan razvoj konvolucijske neuronske mreže primjenjujući VGG arhitekturu u problemu rješavanja klasifikacije slika. Slike pasa i mačaka su zadane iz poznate baze podataka zadane u svrhu Kaggle-ovog natjecanja iz strojnog učenja iz 2013. godine. U uvodu rada se upoznajemo s osnovnom gradivnom jedinicom neuronske mreže perceptorom i osnovnom principima njezinog rada. U nastavku će se predstaviti opća arhitektura konvolucijskih neuronskih mreža. Naglasak ovog rada je na savladavanju principa razvoja osnovnog modela VGG arhitekture neuronske mreže s kojom se želi prikazati rast preciznosti klasifikacije mreže dodavanjem novih blokova. Prikazat će se razvoj osnovnog modela kroz korištenje cloud servisa i online sučelja Google Colab koristeći se otvorenim softverskim bibliotekama za strojno učenje (Tenserflow). Potom se na istoj bazi podataka testira već unaprijed istrenirana arhitektura neuronske mreže VGG-16 te se uspoređuju rezultati s osnovnim modelom. This paper presents the development of a convolutional neural network applying the VGG architecture to the problem of image classification. The images of dogs and cats were taken from a well-known database provided for Kaggle's 2013 machine learning competition. In first part of the paper, we are introduced to the basic building block of the neural network, the preceptor, and the basic principles of its operation. Furthermore, the general architecture of convolutional neural networks will be presented. The emphasis of this paper is on mastering the principles of developing the basic model of the VGG architecture of the neural network, with which we want to show the growth of the accuracy of the network classification by stacking new blocks. The development of the basic model will be presented through the use of cloud services and the Google Colab online interface using open software libraries for machine learning (TensorFlow). Then, the pre-trained VGG-16 neural network architecture is tested on the same database and the results are compared with the basic model.

Zadatak ovog rada je saznati na koji način bi se oblik 3D objekta izmijenio kako bi se poboljšala preciznost određivanja njegove orijentacije. Prikazivanje slika, manipuliranje 3D objektima i potreban okvir za povezivanje 2D slika i 3D objekata je ostvaren bibliotekom PyTorch3D. Opisan je diferencijabilni prikazivač, ekstraktor značajki EfficientNet, konvolucijska neuronska mreža i parametri i funkcije korišteni u funkciji gubitka neuronske mreže. Na kraju su prikazani rezultati testiranja modela i ovisnost izgleda 3D objekta o raznim kombinacijama parametara s naglaskom na funkciju gubitka. The task of this paper is to find out how the shape of a 3D object would change in order to improve the accuracy of determining its orientation. The image generation, manipulation of 3D objects and the necessary framework for connecting 2D images and 3D objects is realized through the PyTorch3D library. A description of the differentiable renderer, EfficientNet feature extractor, convolutional neural network and the parameters and functions used in the neural network loss function is provided. In the end, the paper shows the results of the model and the dependence of the appearance of the 3D object on various combinations of parameters with emphasis on the loss function.

Fire has always exerted a great attraction on humans. Fires generally provide social and environmental impacts at the places where they occur. Several Brazilian localities, especially in the driest months of the year, are more susceptible to this phenomenon. In this paper, an index able of classifying levels of fire risk in areas geographically located in Brazil. This paper presents an index capable of classifying fire risk levels elaborated from neuro-fuzzy systems. Data from the municipality of Sorocaba were used to test the proposed models. The results obtained by this index are promising, reaching values of mean absolute error below 3% when applied in the prediction of the risk of fire for the maximum period of up to 3 days. The proposed index can be used as a tool to support and assist various research agencies or institutes that need to identify the possibility of burning, corroborating the measures to reduce atmospheric emitters and meeting Goal 15 of Agenda 30 as defined by the UN in 2015, which aims to stimulate conservation actions and the recovery and sustainable use of ecosystems. O fogo sempre exerceu grande atração sobre os seres humanos. As queimadas, de maneira geral, proporcionam impactos sociais e ambientais nos locais onde ocorrem. Diversas localidades brasileiras, especialmente nos meses mais secos do ano, estão mais suscetíveis a esse fenômeno. O estudo e o monitoramento do risco do fogo são uma poderosa ferramenta adotada no mapeamento e sensoriamento de áreas afetadas ao longo do território brasileiro e em outras partes do mundo. Este trabalho apresenta um índice para classificar os níveis de risco de fogo, elaborado com base nos sistemas neuro-fuzzy. Dados da cidade de Sorocaba foram utilizados para testar os modelos propostos. Os resultados obtidos mostram-se promissores, alcançando valores referentes à média de erros absolutos abaixo de 3%, aplicados na previsão do risco de queima pelo período máximo de até três dias. O índice proposto poderá ser utilizado como ferramenta de apoio e auxílio a diversos órgãos ou institutos de pesquisa que necessitam identificar a possibilidade de ocorrência de queimadas. Pode, assim, colaborar nas medidas para a redução de emissores atmosféricos, de modo a satisfazer o objetivo 15 da Agenda 30 definido pela Organização das Nações Unidas em 2015, o qual visa estimular ações de conservação, recuperação e uso sustentável de ecossistemas, especialmente.

The production of high-performing thermoelectrics is limited by the high computational energy and time required by the current finite element method solvers that are used to analyze these devices. This paper introduces a new concentrating solar thermoelectric generator made of segmented materials that have non-uniform leg geometry to provide high efficiency. After this, the optimum performance of the device is obtained using the finite element method conducted using ANSYS software. Finally, to solve the high energy and time requirements of the conventional finite element method, the data generated by finite elements are used to train a regressive artificial neural network with 10 neurons in the hidden layer. Results are that the power and efficiency obtained from the optimized device design are 3× and 2× higher than the original unoptimized device design. Furthermore, the developed neural network has a high accuracy of 99.95% in learning the finite element data. Finally, the neural network predicts the modified device performance about 800× faster than the conventional finite element method. Overall, the paper provides insights into how thermoelectric manufacturing companies can harness the power of artificial intelligence to design very high-performing devices while saving time and cost.

This paper introduces explicit neural representations of fundamental hysteresis operators such as the play and stop operators. The hysteresis neurons are represented by recurrent artificial neural networks (ANN) using classical activation functions and are trained with gradient-based learning algorithms. The hysteresis neurons are combined into a single ANN hysteretic layer which, from a mathematical point of view, is equivalent to a classical Prandtl-Ishlinskii model expression, but has all the advantages of ANNs. One such benefit is the flexibility it offers to be combined with other ANN layers to obtain various complex model structures. In this paper, this is illustrated by combining it with a linear recurrent neural network to obtain a Hammerstein model structure, which is often considered for the characterization of systems driven by a class of smart material-based actuators. The input nonlinearity in the Hammerstein model is represented by a Prandtl-Ishlinskii ANN hysteretic layer while the linear dynamics are captured using a linear recurrent ANN representing a single-input multiple-output linear time-invariant state-space model. The identification approach of the Hammerstein model with a hysteretic layer is illustrated both numerically and experimentally.

Problem klasifikacije fragmenata datoteka jedan je od ključnih problema kojim se bavi računalna forenzika. Tehnike koje rješavaju ovaj problem zahtijevaju puno računalnih resursa i vremena kako bi proizvele dobre rezultate. U ovom radu proučava se utjecaj koeficijenata Fourierove transformacije na uspješnost klasifikacije neuronskim mrežama s fokusom na umjetne neuronske mreže. Implementirano je više arhitektura i oformljeno više različitih ulaznih skupova podataka. Dane su usporedbe i komentari rezultata za konstruirane mreže. File fragment classification problem is one of the main problems in computer forensics. The techniques that produce good results are very computationally expensive and time-consuming. This paper studies the impact of Fourier transform coefficients on classification of neural networks, focusing on ANNs. Multiple network architectures were implemented, and multiple sets of input data were formed. The paper compares and comments on the results for each neural network model.