Artificial Neural Network Ensemble Research Articles

Ensemble learning and ground-truth validation of synaptic connectivity inferred from spike trains.

Probing the architecture of neuronal circuits and the principles that underlie their functional organization remains an important challenge of modern neurosciences. This holds true, in particular, for the inference of neuronal connectivity from large-scale extracellular recordings. Despite the popularity of this approach and a number of elaborate methods to reconstruct networks, the degree to which synaptic connections can be reconstructed from spike-train recordings alone remains controversial. Here, we provide a framework to probe and compare connectivity inference algorithms, using a combination of synthetic ground-truth and in vitro data sets, where the connectivity labels were obtained from simultaneous high-density microelectrode array (HD-MEA) and patch-clamp recordings. We find that reconstruction performance critically depends on the regularity of the recorded spontaneous activity, i.e., their dynamical regime, the type of connectivity, and the amount of available spike-train data. We therefore introduce an ensemble artificial neural network (eANN) to improve connectivity inference. We train the eANN on the validated outputs of six established inference algorithms and show how it improves network reconstruction accuracy and robustness. Overall, the eANN demonstrated strong performance across different dynamical regimes, worked well on smaller datasets, and improved the detection of synaptic connectivity, especially inhibitory connections. Results indicated that the eANN also improved the topological characterization of neuronal networks. The presented methodology contributes to advancing the performance of inference algorithms and facilitates our understanding of how neuronal activity relates to synaptic connectivity.

Machine learning predictions for carbon monoxide levels in urban environments

The increasing carbon emissions in Malaysia necessitate accurate methods to track and control pollution levels. This study focuses on predicting carbon monoxide (CO) concentrations in Petaling Jaya using various machine learning models, and two important parameters, CO concentration and time, were considered in the analysis. Six distinct machine learning models were assessed to gauge their predictive capabilities. These models encompassed a range of regression modeling techniques namely; Linear, Tree, Gaussian Process, Ensemble of Trees, Support Vector, and Artificial Neural Networks (ANN) modelling. The Matern 5/2 Gaussian Process Regression (GPR) model consistently outperformed other models across all scenarios, demonstrating high R2 values and low RMSE, MSE, and MAE values. Specifically, in scenarios 1, 2, 3, and 4, the Matern 5/2 model exhibited the lowest RMSE (0.084–0.088) and highest R2 (0.97), highlighting its reliability and robustness in predicting CO concentrations. Additionally, the Rational Quadratic model achieved an R2 of 0.97 with an RMSE of 0.088 in scenario 1, while the Quadratic SVM excelled in scenario 3 with an R2 of 0.965 and low RMSE, MSE, and MAE values (0.085, 0.007, and 0.066). These findings provide valuable insights for environmental protection, renewable energy transition, energy efficiency, sustainable land use planning, and public awareness. However, acknowledging the study's single-area focus and potential limitations in representing diverse regions, further research is warranted to explore carbon emissions across varied areas and enhance the generalizability of the findings.

Comparative Analysis of Ensemble Learning Techniques and ANN for Prediction of Compressive Strength of Concrete Mix

Abstract: In this study, non-linear machine learning regressors and an artificial neural network (ANN) were modeled to predict the compressive strength of concrete mixes. A dataset of 1030 mix design data points with a wide range of compressive strength varying from M10 to M80 is taken into account. The data set comprised of 8 input parameters namely- cement, water, fine & coarse aggregate, flyash, slag, superplasticizer, and age. The paper also analyzes effects on the prediction of compressive strength due to waiving of the parameters with lesser importance factors. To improve the performances of models by reducing the variance, biases, and errors, K-fold cross-validation was examined. To examine the performances of the models, metrics such as coefficient of determination (R2) and mean square error (MSE) were employed. Most of the models can predict the compressive strength with good accuracy. Out of the 7-machine learning regressors and ANN models employed, Gradient Boosting Regressor yielded the best prediction accuracy of 0.9561 with the lowest MSE value of 0.0513.

An ensemble of artificial neural network models to forecast hourly energy demand

AbstractWe propose an ensemble artificial neural network (EANN) methodology for predicting the day ahead energy demand of a district heating operator (DHO). Specifically, at the end of one day, we forecast the energy demand for each of the 24 h of the next day. Our methodology combines three artificial neural network (ANN) models, each capturing a different aspect of the predicted time series. In particular, the outcomes of the three ANN models are combined into a single forecast. This is done using a sequential ordered optimization procedure that establishes the weights of three models in the final output. We validate our EANN methodology using data obtained from a A2A, which is one of the major DHOs in Italy. The data pertains to a major metropolitan area in Northern Italy. We compared the performance of our EANN with the method currently used by the DHO, which is based on multiple linear regression requiring expert intervention. Furthermore, we compared our EANN with the state-of-the-art seasonal autoregressive integrated moving average and Echo State Network models. The results show that our EANN achieves better performance than the other three methods, both in terms of mean absolute percentage error (MAPE) and maximum absolute percentage error. Moreover, we demonstrate that the EANN produces good quality results for longer forecasting horizons. Finally, we note that the EANN is characterised by simplicity, as it requires little tuning of a handful of parameters. This simplicity facilitates its replicability in other cases.

Ensemble of artificial neural networks weighted by operational characteristics for prognosis of acute heart failure

Ensemble of artificial neural networks weighted by operational characteristics for prognosis of acute heart failure

Comparison of individual and ensemble machine learning models for prediction of sulphate levels in untreated and treated Acid Mine Drainage

Machine learning was used to provide data for further evaluation of potential extraction of octathiocane (S8), a commercially useful by-product, from Acid Mine Drainage (AMD) by predicting sulphate levels in an AMD water quality dataset. Individual ML regressor models, namely: Linear Regression (LR), Least Absolute Shrinkage and Selection Operator (LASSO), Ridge (RD), Elastic Net (EN), K-Nearest Neighbours (KNN), Support Vector Regression (SVR), Decision Tree (DT), Extreme Gradient Boosting (XGBoost), Random Forest (RF), Multi-Layer Perceptron Artificial Neural Network (MLP) and Stacking Ensemble (SE-ML) combinations of these models were successfully used to predict sulphate levels. A SE-ML regressor trained on untreated AMD which stacked seven of the best-performing individual models and fed them to a LR meta-learner model was found to be the best-performing model with a Mean Squared Error (MSE) of 0.000011, Mean Absolute Error (MAE) of 0.002617 and R2 of 0.9997. Temperature (°C), Total Dissolved Solids (mg/L) and, importantly, iron (mg/L) were highly correlated to sulphate (mg/L) with iron showing a strong positive linear correlation that indicated dissolved products from pyrite oxidation. Ensemble learning (bagging, boosting and stacking) outperformed individual methods due to their combined predictive accuracies. Surprisingly, when comparing SE-ML that combined all models with SE-ML that combined only the best-performing models, there was only a slight difference in model accuracies which indicated that including bad-performing models in the stack had no adverse effect on its predictive performance.

A Multi-Step Ensemble Approach for Energy Community Day-Ahead Net Load Point and Probabilistic Forecasting

The incorporation of renewable energy systems in the world energy system has been steadily increasing during the last few years. In terms of the building sector, the usual consumers are becoming increasingly prosumers, and the trend is that communities of energy, whose households share produced electricity, will increase in number in the future. Another observed tendency is that the aggregator (the entity that manages the community) trades the net community energy in public energy markets. To accomplish economically good transactions, accurate and reliable forecasts of the day-ahead net energy community must be available. These can be obtained using an ensemble of multi-step shallow artificial neural networks, with prediction intervals obtained by the covariance algorithm. Using real data obtained from a small energy community of four houses located in the southern region of Portugal, one can verify that the deterministic and probabilistic performance of the proposed approach is at least similar, typically better than using complex, deep models.

Enhancing lung cancer disease diagnosis by employing ensemble deep learning approaches

<span>Cancer is a disease that results from the unnatural proliferation of aberrant cells that infest the body’s healthy cells and spread throughout the body. Lung cancer is characterized by an imbalance in the cells of the affected organs, namely the lungs. The prediction of lung cancer at an early stage is very important, particularly in countries that are densely populated and have lower incomes. Clinically conventional approaches, such as blood tests and other types of treatments, are used by specialists. The age of artificial intelligence (AI) has begun, and today, it is feasible to construct a computer-aided diagnostic mechanism with the assistance of machine learning and deep learning algorithms. In this particular piece of research, one deep learning algorithm, an artificial neural network (ANN), has been investigated to determine whether or not lung cancer could be detected at an earlier stage. In addition to conventional ANN, ensemble ANN with weighted averaging and soft and hard voting ensemble techniques are also considered. In order to achieve this effectiveness, the state-of-the-art parameters for the proposed method using ANN are assessed and evaluated using the lung cancer dataset. The empirical analysis shows that hard voting-enabled ANN shows the highest accuracy at 97.47%.</span>

Early Fast Cost Estimates of Sewerage Projects Construction Costs Based on Ensembles of Neural Networks

This paper presents research results on the development of an original cost prediction model for construction costs in sewerage projects. The focus is placed on fast cost estimates applicable in the early stages of a project, based on fundamental information available during the initial design phase of sanitary sewers prior to the detailed design. The originality and novelty of this research lie in the application of artificial neural network ensembles, which include a combination of several individual neural networks and the use of simple averaging and generalized averaging approaches. The research resulted in the development of two ensemble-based models, including five neural networks that were trained and tested using data collected from 125 sewerage projects completed in the Czech Republic between 2018 and 2022. The data included information relevant to various aspects of projects and contract costs, updated to account for changes in costs over time. The developed models present satisfactory predictive performance, especially the ensemble model based on simple averaging, which offers prediction accuracy within the range of ±30% (in terms of percentage errors) for over 90% of the training and testing samples. The developed models, based on the ensembles of neural networks, outperformed the benchmark model based on the classical approach and the use of multiple linear regression.

Short-lead seasonal precipitation forecast in northeastern Brazil using an ensemble of artificial neural networks

This study assesses the deterministic and probabilistic forecasting skill of a 1-month-lead ensemble of Artificial Neural Networks (EANN) based on low-frequency climate oscillation indices. The predictand is the February-April (FMA) rainfall in the Brazilian state of Ceará, which is a prominent subject in climate forecasting studies due to its high seasonal predictability. Additionally, the study proposes combining the EANN with dynamical models into a hybrid multi-model ensemble (MME). The forecast verification is carried out through a leave-one-out cross-validation based on 40 years of data. The EANN forecasting skill is compared with traditional statistical models and the dynamical models that compose Ceará’s operational seasonal forecasting system. A spatial comparison showed that the EANN was among the models with the smallest Root Mean Squared Error (RMSE) and Ranked Probability Score (RPS) in most regions. Moreover, the analysis of the area-aggregated reliability showed that the EANN is better calibrated than the individual dynamical models and has better resolution than Multinomial Logistic Regression for above-normal (AN) and below-normal (BN) categories. It is also shown that combining the EANN and dynamical models into a hybrid MME reduces the overconfidence of the extreme categories observed in a dynamically-based MME, improving the reliability of the forecasting system.

5-lead 3D-vectorcardiography differentiates between high and low cardiovascular risk profiles in patients with suspected or know coronary heart disease

Abstract 5-lead 3D-vectorcardiography (3D-VCG), including a dorsal electrode with ensemble of artificial neural networks (ANN) offers additional information over standard 12-lead electrocardiography (ECG) in the detection of cardiac ischaemia. Artificial Intelligence (AI) and access to a massive global clinical data repository have the potential to boost the performance of VCG beyond conventional techniques. VCG offers ease-of-use at rest and is therefore a promising candidate for becoming a new screening tool for coronary artery diseases (CAD). We, thus, tested the hypothesis of 3D-VCG being able to differentiate between patients with lower and high risk of cardiovascular disease (CVD) and identify those requiring coronary interventions. In this exploratory study, we analysed patients in a national multicentre trial. Patients were recruited from general cardiologists’ praxis and radiologist centre with patients referred for further diagnosis of suspected or confirmed CVD. Based on the Perfusion (P)-Factor of the 5-lead 3D- vectorcardiography, including 731 parameters, e.g., QRS-T angle and in-house features calculated in time and frequency domains, such as beat moments a P-Index is calculated. Based on the P-Index, patients were either classified as high, medium, or low risk for CVD (medium + high defined as high CVD-risk). Confirmation of CAD was performed according to the practitioners’ discretion blinded to the P-Index. Number of risk factors (mod. PROCAM score) were compared between the high- and low-risk group using an independent t-test. The 468 patients in this sample (m:w 61:39, age: 66 [40-87] years) showed 3.6 [0 - 7] CVRF, 16% had arrhythmias or conduction disturbances (AF, PM, BBB), 24% pts. had consecutive PCI or CABG. 62% were classified as low while 38% were classified as high risk for CVD by the P-Index. Number of cardiovascular risk factors was significantly higher in the high-risk CVD-group as defined by P-Index compared to low CVD-risk (4.0 [3.0 – 5.0] vs. 3.5 [2.0 – 4.0], p < 0.05). P-Index differentiated between suspected CVD with or without consequent PCI or CABG (Chi2 = 4.02, p<0.05). The AI-based 3D-VCG is an innovative diagnostic tool that can help determine a patient's cardiovascular risk profile in resting condition for clinical and research purposes. Furthermore, the P-Index identified those with significant stenoses in need for coronary intervention and may thus be a further scalable screening method to differentiate patients with consequent interventional approach or need for risk modification. These preliminary data will have to be confirmed in a prospective, large-scale performance study to verify the diagnostic accuracy.

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

Building energy demand accounts for an important part of global energy consumption and emissions. Load forecasting is a crucial functionality for building management systems (BMSs) in order to optimize the energy supply and use with the goal of achieving economic and environmental benefits. Furthermore, if the building is a sports/event venue (one of the most energy-intensive facility type), load forecasting acquires an even greater importance.This paper presents a new forecasting hybrid technique that combines an ensemble of Artificial Neural Networks (ANNs) with an AutoRegressive Integrated Moving Average with eXternal inputs (ARIMAX) model. The proposed methodology is designed for the 24-hour-ahead electrical consumption forecasting of sports venues. These types of buildings are generally characterized by large fluctuations in their load caused by the hosted events, and therefore they are challenging for forecasting tools. The ANNs ensemble is composed of the average output of several multi-layer perceptrons, while the ARIMAX is used to correct the prediction provided by the ANNs working on the ensemble residual. The proposed approach can be easily implemented within a BMS that can be crucial for an efficient facility management. The hybrid methodology has been tested and validated on the load data of a real football arena. The proposed methodology achieves a MAPE of 9.03 % on an open dataset, and 9.38% on a closed dataset, outperforming standard machine learning techniques, hybridization techniques involving time-series and machine learning methods, as well as a state of the art event venue load forecasting technique. Furthermore, a feature importance study reveals that all the selected inputs are crucial in achieving the result. Finally, the inclusion of results related to an open dataset encourages further studies and comparisons.

Robust Design of Herringbone Grooved Journal Bearings Using Multi-Objective Optimization With Artificial Neural Networks

Abstract Herringbone grooved journal bearings (HGJBs) are widely used in micro-turbocompressor applications due to their high load-carrying capacity, low friction, and oil-free solution. However, the performance of these bearings is sensitive to manufacturing deviations, which can lead to significant variations in their performance and stability. In this study, design guidelines for robust design against manufacturing deviations of HGJB supported micro-turbocompressors are proposed. These guidelines are based on surrogate model-assisted multi-objective optimization using ensembles of artificial neural networks trained on a large dataset of rotor and bearing designs as well as operating conditions. The developed framework is then applied to a series of case studies representative of heat-pump and fuel-cell micro-turbomachines. To highlight the importance of rotor geometry and bearing aspect ratio in the robustness of HGJBs, two types of optimizations are performed: one focusing on optimizing the bearing geometry, and the other focusing on both the bearing and rotor geometries. The analysis of the Pareto fronts and Pareto optima of each type of optimization and case study allows for the derivation of design guidelines for the robust design of HGJB supported rotors. Results suggest that by following these guidelines, it is possible to significantly improve the robustness of herringbone grooved journal bearings against manufacturing deviations, resulting in stable operation. The best design achieved ±8 μm tolerance on the bearing clearance, and designs optimized for both rotor and bearing geometry outperformed those optimized for bearing geometry alone. This work successfully identifies guidelines for the robust design of herringbone grooved journal bearings in micro-turbocompressor applications, demonstrating the strength of surrogate model-assisted multi-objective optimization. It provides a valuable tool for engineers seeking to optimize the performance and reliability of these bearings.

Ensemble of Artificial Neural Networks for Approximating the Survival Signature of Critical Infrastructures

Abstract Survival signature can be useful for the reliability assessment of critical infrastructures. However, analytical calculation and Monte Carlo Simulation (MCS) are not feasible for approximating the survival signature of large infrastructures, because of the complexity and computational demand due to the large number of components. In this case, efficient and accurate approximations are sought. In this paper we formulate the survival signature approximation problem as a missing data problem. An ensemble of artificial neural networks (ANNs) is trained on a set of survival signatures obtained by MCS. The ensemble of trained ANNs is, then, used to retrieve the missing values of the survival signature. A numerical example is worked out and recommendations are given to design the ensemble of ANNs for large-scale, real-world infrastructures. The electricity grid of Great Britain, the New England power grid (IEEE 39-Bus Case), the reduced Berlin metro system and the approximated American Power System (IEEE 118-Bus Case) are, then, eventually, analyzed as particular case studies.

Data-Driven Model to Predict Aircraft Vibration Environment

Vibration levels that onboard equipment must be able to withstand throughout their lives for correct operation are mainly determined experimentally because predicting the dynamic behavior of a complete aircraft requires computational means and methods that are currently difficult to implement. We present a data-driven methodology that leverages flight-test accelerometer data to produce a predictive model. This model, based on an ensemble of artificial neural networks, performs a multioutput multivariate regression to estimate vibration spectra from a set of aircraft general parameters without having to characterize excitation sources. The model is compared with baseline models over two protocols, which are 1) standard training and testing as well as 2) extrapolation to high dynamic pressures, in order to assess physical consistency. Although the first protocol shows that all models can produce results accurate enough for this context, the second protocol shows that only the ensemble model is able to correctly extrapolate the energy. Using the Shapley additive explanations method, also known as SHAP, we show that these results can be explained by the ability of our model to identify the dynamic pressure as the core feature used in the extrapolation protocol. The proposed model can be used in multiple applications, such as anomaly detection and vibration flight envelope opening.

Utilizing different machine learning methods to accurately predict density, temperature, velocity, and thermal conductivity of hydrophilic, hydrophobic, and compound materials

A learning algorithm is a set of instructions used in machine learning that enables a computer program to imitate how humans improve at describing specific types of data. As programming is exposed to more data, math and logic that support a learning algorithm can be updated (without human intervention) over time. In this study, different machine learning regression methods were used to predict the thermal conductivity (knf) -related physical properties of tested samples. Different parameters were considered such as density, temperature, rate, velocity, and knf is the most significant to improve heat transfer. Gaussian regression, linear regression, support vector regression, generalized linear regression, regression tree, artificial neural network (ANN), LSBoost Ensemble, and bag Ensemble were used. Gaussian regression method was produced the best results based on research comparing these methods. This method produced more accurate results in four fields (density, temperature, rate, and knf) than other methods. However, the best results in velocity field via LSBoost regression, which entails calculating the RMSE and PLCC. After selecting the optimal regression, the results of this regression should be validated. All of trained models had a strong correlation with real-world values, based on the findings of study, which are positive. Finally, it was determined that the covariance function was the only parameter that affected the RMSE results in Gaussian regression. This study illustrates the effect of using distinct covariance functions for each RMSE physical properties. Density is expressed over exponential function, and velocity and rate are expressed in mother32. The temperature was expressed due to the exponential square, and knf was expressed in terms of exponential.

Combination of deep learning and ensemble machine learning using intraoperative video images strongly predicts recovery of urinary continence after robot-assisted radical prostatectomy.

We recently reported the importance of deep learning (DL) of pelvic magnetic resonance imaging in predicting the degree of urinary incontinence (UI) following robot-assisted radical prostatectomy (RARP). However, our results were limited because the prediction accuracy was approximately 70%. To develop a more precise prediction model that can inform patients about UI recovery post-RARP surgery using a DL model based on intraoperative video images. The study cohort comprised of 101 patients with localized prostate cancer undergoing RARP. Three snapshots from intraoperative video recordings showing the pelvic cavity (prior to bladder neck incision, immediately following prostate removal, and after vesicourethral anastomosis) were evaluated, including pre- and intraoperative parameters. We evaluated the DL model plus simple or ensemble machine learning (ML), and the area under the receiver operating characteristic curve (AUC) was analyzed through sensitivity and specificity. Of 101, 64 and 37 patients demonstrated "early continence (using 0 or 1 safety pad at 3 months post-RARP)" and "late continence (others)," respectively, at 3 months postoperatively. The combination of DL and simple ML using intraoperative video snapshots with clinicopathological parameters had a notably high performance (AUC, 0.683-0.749) to predict early recovery from UI after surgery. Furthermore, combining DL with ensemble artificial neural network using intraoperative video snapshots had the highest performance (AUC, 0.882; sensitivity, 92.2%; specificity, 78.4%; overall accuracy, 85.3%) to predict early recovery from post-RARP incontinence, with similar results by internal validation. The addition of clinicopathological parameters showed no additive effects for each analysis using DL, EL and simple ML. Our findings suggest that the DL algorithm with intraoperative video imaging is a reliable method for informing patients about the severity of their recovery from UI after RARP, although it is not clear if our methods are reproducible for predicting long-term UI and pad-free continence.

COMPARATIVE ANALYSIS OF THE EFFECTIVENESSOF CORRELATION-REGRESSION AND NEURAL NETWORK MODELING IN PREDICTING ENERGY EMISSIONSOF CARBON DIOXIDE IN RUSSIA

Effective national cap-and-trade system involves accurate projections of greenhouse gas emissions for the national economy as a whole and by industry. The main source of carbon dioxide emissions in most countries of the world (including Russia) is the energy sector with 9traditional fuels (coal, gas and oil). The objective of the paper is to forecast energy emissionsof carbon dioxide in the Russian Federation by applying adequate economic and mathematical modelling methods. To achieve it, two hypotheses are consistently put forward and tested: the possibility of building a medium-term forecast of the indicator as a result of correlation and regression analysis and the one based on the formation of a Bayesian ensemble of artificial neural networks. Both hypotheses are confirmed in the empirical study. However, the second method provides a higher degree of accuracy in approximating statistical data. Therefore, within the framework of this article, the formation of medium-term forecasts of energy carbondioxide emissions in Russia is made with the help of neural network modeling. Highly accurate forecasting provides a scientific basis for effective policymakers' decisions in decarbonisation of the national economy.

Parametric analysis and machine learning for enhanced recovery of high-value sugar from date fruits using supercritical CO2 with co-solvents

The extraction of date sugar using supercritical extraction is a process that is still in its formative stages. In this study, a comprehensive parametric analysis of the supercritical fluid extraction (SFE) process using supercritical CO2 with water/ethanol as co-solvents was performed to achieve maximum recovery of date sugar extract. The results showed that the maximum total sugar content (TSC) was 70.45 ± 0.01 g/100 g of DFP. This was made up of 7.42 g/100 g fructose, 6.49 g/100 g glucose, and 56.54 g/100 g sucrose. This was attained with 15 v/v% water as co-solvent, 50 ℃, and 200 bar. In addition, machine learning with non-linear regression and artificial neural network (ANN) ensembles was used for TSC prediction. The ANN results showed a strong correlation between operating parameters and sugar recovery with a total R2 of 0.986 ± 0.010. Compared to conventional hot water extraction method (CHWE), the CO2-SFE process resulted in a 1.4-fold increase in TSC recovery and a 2.1-fold increase in organic acids recovery. CO2-SFE demonstrated comparable TSC results with a difference of only 1.2% when compared to the ultrasound-assisted extraction ‘'USAE’ method. The results of the detailed chemical analysis (HPLC and FT-IR) and morphological analysis (SEM) showed that the USAE and CO2-SFE were more efficient than CHWE. Supercritical extraction with co-solvents is particularly effective in recovering date sugar from date fruit, making it a desirable ingredient in a variety of food products.

Consensus Ensemble Multitarget Neural Network Model of Anxiolytic Activity of Chemical Compounds and Its Use for Multitarget Pharmacophore Design.

A classification consensus ensemble multitarget neural network model of the dependence of the anxiolytic activity of chemical compounds on the energy of their docking in 17 biotargets was developed. The training set included compounds thathadalready been tested for anxiolytic activity and were structurally similar to the 15 studied nitrogen-containing heterocyclic chemotypes. Seventeen biotargets relevant to anxiolytic activity were selected, taking into account the possible effect on them of the derivatives of these chemotypes. The generated model consistedof three ensembles of artificial neural networks for predicting three levels of anxiolytic activity, with sevenneural networks in each ensemble. A sensitive analysis of neurons in an ensemble of neural networks for a high level of activity made it possible to identify four biotargets ADRA1B, ADRA2A, AGTR1, and NMDA-Glut, which were the most significant for the manifestation of the anxiolytic effect. For these four key biotargets for 2,3,4,5-tetrahydro-11H-[1,3]diazepino[1,2-a]benzimidazole and [1,2,4]triazolo[3,4-a][2,3]benzodiazepine derivatives, eight monotarget pharmacophores of high anxiolytic activity were built. Superposition of monotarget pharmacophores built two multitarget pharmacophores of high anxiolytic activity, reflecting the universal features of interaction 2,3,4,5-tetrahydro-11H-[1,3]diazepino[1,2-a]benzimidazole and [1,2,4]triazolo[3,4-a][2,3]benzodiazepine derivatives with the most significant biotargets ADRA1B, ADRA2A, AGTR1, and NMDA-Glut.