Artificial Neural Network Research Articles

Predicting uniaxial compressive strength of tuff after accelerated freeze-thaw testing: Comparative analysis of regression models and artificial neural networks

Predicting uniaxial compressive strength of tuff after accelerated freeze-thaw testing: Comparative analysis of regression models and artificial neural networks

Artificial neural network models for predicting breaking strength and abrasion resistance properties of woven fabrics with different chenille yarns

Artificial neural network models for predicting breaking strength and abrasion resistance properties of woven fabrics with different chenille yarns

Anxiety in children and adolescents with autism spectrum disorder: behavioural phenotypes and environmental factors.

Anxiety is the most prevalent comorbidity among children and adolescents with autism spectrum disorder (ASD), yet little is known about the associated risk factors. In a heterogenous cohort of children aged 5-18years old (n = 262, 42% ASD), participants and their parents completed standardized questionnaires to assess anxiety, ASD symptom severity, inattention/hyperactivity, emotional problems, depressive symptoms, parental styles and stress, and demographic factors. An artificial neural network analysis using a self-organizing map, a statistical technique used to cluster large datasets, revealed 3 distinct anxiety profiles: low (n = 114, 5% ASD), moderate (n = 70, 64% ASD) and high (n = 78, 96% ASD) anxiety. A recursive feature elimination analysis revealed that depression and peer problems contributed the most to differences between the anxiety profiles. Difficulties with peers in individuals with ASD who experience anxiety may be related to challenges with social competence and this may heighten depressive symptoms. Findings highlight the importance of assessing depressive symptoms in children and adolescents with ASD who experience anxiety. Identifying anxiety profiles among children and adolescents with ASD may prove beneficial in clinical practice by facilitating the development of tailored interventions that aid in managing anxiety and depressive symptoms. Furthermore, strengthening social communication skills may improve peer relationships and could aid in managing depressive symptoms among children and adolescents with ASD who experience anxiety.

AI-based modeling studies for dye removal using mixed biomass composites from algae and plant seeds: Isotherm, kinetics, and mechanistic insights

Biomass based remediation have emerged as a promising solution for the economic and sustainable pollutant removal from environment. In this study, novel mixed biomass derived from algae and manila tamarind seeds was synthesized and extensively characterized for application in dye adsorption. Mesoporous structure with pore diameter of 4.006 nm was observed from Brunauer-Emmett-Teller (BET) characterization. Best fit for isotherm was identified with Langmuir model indicating the monolayer nature of dye removal. Artificial intelligence (AI) based models – artificial neural network (ANN) and adaptive neuro fuzzy interference (ANFIS) system were employed for the prediction of dye removal under different conditions, a unique approach which has not been explored in previous studies based on mixed biomass adsorption. In addition, embedding particle swarm optimization (PSO) had a significant improvement in prediction of dye removal. For both Eriochrome black (EB) dye and Brilliant Orange (BO) dye adsorption employing mixed biomass, the ANN-PSO model yielded the best correlation of 0.9999 and 0.9994. For eriochrome black and brilliant orange dye, the maximum adsorption of 79.95 mg/g and 102.4 mg/g for the mixed biomass composite was noted, highlighting the adsorption efficiency of biomass. The uniqueness of the study lies in the integration of mixed biomass adsorption with AI driven modeling, providing a dual approach to enhance dye removal and adsorption efficiency. In addition, the regenerated mixed biomass can be reused for five consecutive cycles, highlighting its practical application, distinguishing from existing research that focused on single use biomass systems.

Improving the coverage area and flake size of ReS2 through machine learning in APCVD

Machine learning is playing a crucial role in optimizing material synthesis, particularly in scenarios where several parameters related to growth exhibit different and significant outcomes. An example of such complexity is the growth of atomically thin semiconductors through chemical vapor deposition (CVD), where multiple parameters can influence the thermodynamics and reaction kinetics involved in the synthesis. Herein, we performed a set of orthogonal experiments, varying the key parameters such as temperature, carries gas flux and precursor position to identify the optimal conditions for maximizing covered area and the size of rhenium disulfide (ReS2) crystals. The experimental results were used to establish correlations among the three thermodynamic variables through an artificial neural network. Contour plots were then generated to visualize the impact on the coverage and flake size of the crystals. This study demonstrates the capability of machine learning to enhance the potential of CVD-growth for the integration of 2D semiconductors like ReS2at larger scales.

Texture-Based Classification of Fetal Growth Restriction From Intrauterine Neurosonographic Image.

Fetal growth restriction (FGR) is a condition where fetuses fail to reach their genetic potential for growth, posing a significant health challenge for newborns. The aim of this research was to explore the efficacy of texture-based analysis of neurosonographic images in identifying FGR in fetuses, which may provide a promising tool for early assessment of FGR. A retrospective analysis collected 100 intrauterine neurosonographic images from 50 FGR and 50 gestational age-appropriate fetuses. Using MaZda software, approximately 300 texture features were extracted from occipital white matter (OWM) and cerebellum of intrauterine neurosonographic images, respectively. Then 10 optimal features were separately selected by 3 algorithms, including the Fisher coefficient method, the method of minimizing classification error probability and average correlation coefficients, and the mutual information coefficient method. Further, the 10 statistically most significant features were selected from these sets to form the mixed feature set. After nonlinear discriminant analysis was performed to reduce feature dimensionality, the artificial neural network (ANN) classifier was conducted, respectively. For OWM and cerebellum, a total of 11 and 14 statistically significant features were selected. When the mixed feature sets of OWM and cerebellum were applied to ANN classifier, classification accuracy were 90.00% (κ = 0.800; P < .001) and 93.00% (κ = 0.860; P < .001), and the receiver operating characteristic curve for identifying FGR showed an area under the curve of 0.82 and 0.87. Texture analysis of fetal intrauterine neurosonographic images is a feasible and noninvasive strategy for evaluating FGR fetuses.

Optimization of reed-based polyol production driven by response surface methodology and machine learning: Process modeling and hydroxyl value prediction

In pursuit of sustainable development goals, the transformation of biomass resources into high-value chemicals has become a focal point within the academic community. Despite being a biomass resource, reed (Phragmites australis) has not yet fully realized its potential in the production of bio-based polyols. This study utilized Response Surface Methodology (RSM) and an Artificial Neural Network (ANN) enhanced by a Genetic Algorithm (GA) to model and optimize the liquefaction process of reeds. Furthermore, this study employed a pre-existing predictive model to estimate the hydroxyl value of reed-derived liquefaction products, thereby aiming to curtail expenses during the biomass chemical development phase. The results demonstrated that both the RSM and GA-ANN models accurately predict bio-polyol yield, with the RSM model outperforming the ANN model. Based on optimal conditions predicted by the RSM model, the best experimental process parameters were established: raw material particle size 2–12 mesh, solid-liquid ratio 5:1, glycerol content 32.5 %, catalyst content 3.6 %, temperature 169 °C, and reaction time 58 minutes. Under these conditions, the polyol yield reached 89.145 %, with a relative deviation of 2.366 %, and the bio-polyol exhibited a viscosity of 0.51 Pa·s. The predicted hydroxyl value for the bio-based polyol was found to be 399.19 mg KOH/g. The liquefied product is rich in hydroxyl groups, indicating its potential application value in preparing bio-based polymer materials. This study introduces innovative approaches to the economical production of bio-based polyols at the laboratory scale, which may pave the way for their broader industrial implementation.

New forest fire assessment model based on artificial neural network and analytic hierarchy process or fuzzy-analytic hierarchy process methodology for fire vulnerability map

Forest fires significantly disrupt global ecosystems. Many forecasting techniques predict fire activity and allocate prevention resources, but various factors are missing from the assessments, and multi-criteria decision approaches alone are insufficient. This work introduced a novel methodology combining artificial neural networks (ANN) with the analytical hierarchy process (AHP), fuzzy AHP multi-criteria methods, and spatial data to detect potential forest fire vulnerability areas using twenty variables. The results from AHP or fuzzy AHP, were processed using a multilayer perceptron with a backpropagation algorithm. The final ANN model has two objectives: first, to create a forest fire vulnerability with four classes (low, moderate, high, and very high), and second, to classify burned area sizes in regions highly or very highly vulnerable to fire. Evaluation metrics were also applied for validation. The fire model was tested using both literature review data and in situ observations. Comparative analysis showed that the burned area size model performed better than other machine learning methods, achieving an accuracy score of 89%. Meanwhile, the fire vulnerability model scored 82%. The study addresses the problem of prediction and provides an algorithm for classifying fire risk based on historical data in the Czech Republic, offering a master model for future forest management. Therefore, the ANN model, once trained, validated, and tested, does not require resetting and is effective for estimating forest fire vulnerability. Moreover, it produces results quickly, providing rapid insights into complex forest ecosystems, saving time and enhancing understanding for decision-makers.

A novel interleaved BIFRED-Zeta with intelligent MPPT for PV applications

ABSTRACT The current energy perspective postulates profound presence of Renewable Energy Sources (RESs), resulting in need for optimal DC-DC converter design to assure efficient power conversion and minimise losses. By minimising energy losses and improving power quality, an optimal converter design helps ensure a stable and reliable power supply from RESs, which is essential for meeting the growing demand for clean energy. Therefore, this research introduces an innovative Interleaved BIFRED-Zeta converter configuration, designed to meet all criteria associated with an optimal converter design. The interleaved architecture of converter allows the reduction of current ripples and voltage stress along with the improvement of efficiency. Furthermore, the absence of transformer in converter structure enables it to be simple, cost effective and less bulky. The Interleaved Boost Integrated Flyback Rectifier Energy Storage DC-DC (BIFRED)-Zeta converter duty cycle is adjusted continuously using Cascaded Artificial Neural Network (ANN) Maximum Power Point Tracking (MPPT) to further maximise Photovoltaics (PV) system power output. The capability of suggested converter design is demonstrated by the results obtained using simulation and laboratory prototype implementation. Thereby, a peak efficiency of 97.6% is estimated to be achieved using suggested converter design.

A (2+1) modified KdV equation with time-dependent coefficients: exploring soliton solution via Darboux transformation and artificial neural network approach

A (2+1) modified KdV equation with time-dependent coefficients: exploring soliton solution via Darboux transformation and artificial neural network approach

Device packaging and integration optimization based on neural network method: Effect of microchannel structure on heat sink performance

With the continuous improvement of power density of electronic device packaging, microchannel structure based on advanced packaging integration technology has become one of the hot topics in thermal management research. In this paper, two heat sink models are numerically simulated: rectangular microchannel and microchannel-pinfin hybrid. The effects of microchannel structure, inlet/outlet arrangement and pinfin design on heat sink performance were studied. The heat sink performance is evaluated by the maximum temperature at hotspot area, the total thermal resistance and the mean absolute temperature deviation (MATD). The research results show that at the same Reynolds number, the hybrid structure with pinfin has better heat sink performance. Compared with the rectangular microchannel structure, the maximum temperature at the hotspot area is reduced by 20.45 K and the total thermal resistance is reduced by 5.9 %. Compared with the three inlet/outlet arrangements of I-type, U-type and Z-type, the I-type hybrid structure has the best heat sink performance. The average absolute temperature deviation is about 10 K at high Reynolds number, and the temperature distribution is more uniform. In addition, the heat sink performance is improved by increasing the pinfin size to expand the heat transfer area. This paper uses simulation results to build an artificial neural network model in order to quickly and accurately predict numerical results under different parameter conditions. The network model has good accuracy and robustness, with a mean absolute error of 0.376 %. The prediction model takes heat source power, Reynolds number and pinfin structural parameters as input. Rapidly predicting the performance of heat sinks with different structural parameters under different boundary conditions will effectively save experimental and computational costs. This research provides new ideas and tools for optimizing heat sink design, and helps improve the application of microchannel cooling technology in electronic devices.

A method for temperature sensor and model selection for machine tool thermal error modelling using ANFIS and ANN

AbstractThermal errors account for a significant part of the dimensional errors of components produced by precision machine tools. These errors are commonly compensated using predictions from temperature-based empirical models. The accuracy and robustness of these predictions are affected by the locations of temperature sensors used to obtain the model’s input data. Methods for sensor selection found in literature are often difficult to replicate and automate because they require tuning of several hyperparameters. This work presents a sensor and model selection approach that uses proper orthogonal decomposition (POD) and QR pivoting to select a subset of sensors that have been preinstalled in the machine tool as possible model inputs. These sensors are then sorted according to their correlation with the thermal error being modelled. The final set of inputs and thermal error model structure is chosen using Bayesian Information Criterion (BIC) to limit model overfitting. The approach was tested by modelling the Y-axis thermal error measured from air-cutting experiments performed under different spindle speeds and feed rates. This enabled determination of the structure and the choice inputs for Adaptive Neuro Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN) thermal error models. The accuracy of these models was compared to that of models trained using inputs selected by conventional approaches: the least absolute shrinkage and selection operator (LASSO), fuzzy c-means clustering (FCM), and principal component regression (PCR). The presented approach had better or comparable results to the conventional approaches while using fewer inputs. The presented approach is also well suited for automation compared to conventional approaches that require expert input.

Exploring the origins of switching dynamics in a multifunctional reservoir computer

The concept of multifunctionality has enabled reservoir computers (RCs), a type of dynamical system that is typically realized as an artificial neural network, to reconstruct multiple attractors simultaneously using the same set of trained weights. However, there are many additional phenomena that arise when training a RC to reconstruct more than one attractor. Previous studies have found that in certain cases, if the RC fails to reconstruct a coexistence of attractors, then it exhibits a form of metastability, whereby, without any external input, the state of the RC switches between different modes of behavior that resemble the properties of the attractors it failed to reconstruct. In this paper, we explore the origins of these switching dynamics in a paradigmatic setting via the “seeing double” problem.

An artificial neural network based approach for harmonic component prediction in a distribution line

With the increasing use of nonlinear devices in both generation and consumption of power, it is essential that we develop accurate and quick control for active filters to suppress harmonics. Time delays between input and output are catastrophic for such filters which rely on real-time operation. Artificial Neural Networks (ANNs) are capable of modeling complex nonlinear systems through adjustments in their learned parameters. Once properly trained, they can produce highly accurate predictions at an instantaneous time frame. Leveraging these qualities, various complex control systems may be replaced or aided by neural networks to provide quick and precise responses. This paper proposes an ANN-based approach for the prediction of individual harmonic components using minimal inputs. By extracting and analyzing the nature of harmonic component magnitudes obtained from the survey of a particular area through real-time measurements, a sequential pattern in their occurrence is observed. Various neural network architectures are trained using the collected data and their performances are evaluated. The best-performing model, whose losses are minimal, is then used to observe the harmonic cancellation for multiple unseen cases through a simplified simulation in hardware-in-the-loop. These neural network structures, which produce instantaneous and accurate outputs, are effective in harmonic filtering.

Development of ANN model for predicting mechanical properties of 3D printed PEEK polymer using FDM and optimization of process parameters for better mechanical properties

Additive manufacturing (AM) has evolved from a proven technology to a rapid prototyping tool with great potential. This technology is widely used in many industries such as automotive, aerospace, and medical; fused deposition modeling (FDM) is a popular 3D printing technique for producing PEEK (polyether ether ketone) parts, including implant prosthetic teeth. In this study, artificial neural network (ANN) modeling, parametric optimization, and experimental examination of PEEK 3D printing were conducted to enhance 3D printing processes. In this study, four critical process factors (infill density, layer height, printing speed, and infill pattern) influence the surface roughness, mechanical strength, and elastic modulus of the printed samples. Utilizing a 4–12–3 network design, this study demonstrates that an ANN model with an average error of less than 5% is optimal for all three responses. Furthermore, the study employed a teaching and learning based optimization algorithm (TLBO) and a non-dominated sorting genetic algorithm (NSGA) to optimize the printing process to obtain improved mechanical properties. The findings highlight TLBO’s ability to minimize surface roughness to 6.01 μm and NSGA’s capability to maximize the elastic modulus to 1253.35 MPa and ultimate tensile strength to 65.55 MPa. Microstructural studies supported the results obtained through parametric analysis and optimization.

Multi-omics analysis and response prediction of PD-1 monoclonal antibody containing regimens in patients with relapsed/refractory diffuse large B-cell lymphoma

Patients with relapsed/refractory (r/r) diffuse large B-cell lymphoma (DLBCL) show varied responses to PD-1 monoclonal antibody (mAb) containing regimens. The mechanisms and predictive biomarkers for the efficacy of this regimen are unclear. This study retrospectively collected r/r DLBCL patients who received PD-1 mAb and rituximab regimens as salvage therapy. Clinical and genomic features were collected, and mechanisms were explored by multiplex immunofluorescence and digital spatial profiling. An artificial neural network (ANN) model was constructed to predict the response. Between October 16th, 2018 and May 4th, 2023, 50 r/r DLBCL patients were collected, 29 were response patients and 21 were non-response patients. CREBBP (p = 0.029) and TP53 (p = 0.015) alterations were statistically higher in non-response patients. Patients with PD-L1 CPS ≥ 5 were correlated with a longer overall survival (OS) than those with PD-L1 CPS < 5 (median OS: not reached vs. 9.7 months, hazard ratio [HR]: 3.8, 95% confidence interval [CI] 0.64–22.44, p = 0.016). Immune-related pathways were activated in response patients. The proportion and spatial organization of tumor-infiltrating immune cells affect the response. PD-L1 CPS level, age, and alterations of TP53, MYD88, CREBBP, EP300, GNA13 were used to build an ANN predictive model that showed high prediction efficiency (training set area under curve [AUC] of 0.97 and test set AUC of 0.94). The proportion and spatial distribution of tumor-infiltrating immune cells may be related to the function of immune-related pathways, thereby influencing the efficacy of PD-1 mAb containing regimens. The ANN predictive model showed potential value in predicting the responses of r/r DLBCL patients received PD-1 mAb and rituximab regimens.

Ultrafast Silicon/Graphene Optical Nonlinear Activator for Neuromorphic Computing

AbstractOptical neural networks (ONNs) have shown great promise in overcoming the speed and efficiency bottlenecks of artificial neural networks. However, the absence of high‐speed, energy‐efficient nonlinear activators significantly impedes the advancement of ONNs and their extension to ultrafast application scenarios like real‐time intelligent signal processing. In this work, a novel silicon/graphene ultrafast all‐optical nonlinear activator, leveraging the hybrid integration of silicon slot waveguides, plasmonic slot waveguides, and monolayer graphene is demonstrated. Exploiting the exceptional picosecond‐scale photogenerated carrier relaxation time of graphene, the response time of the activator is markedly reduced to ≈93.6 ps, establishing all‐optical activator as the fastest known in silicon photonics to knowledge. Moreover, the all‐optical nonlinear activator holds a low threshold power of 5.49 mW and a corresponding power consumption per activation of 0.51 pJ. Its feasibility and capability for use in ONNs, manifesting performance comparable with commonly used activation functions are experimentally confirmed. This breakthrough in speed and energy efficiency of all‐optical nonlinear activators opens the door to significant improvements in the performance and applicability of ONNs.

On the construction of an artificial neural network for solving a system of equations Navier–Stokes in the case of incompressible fluid

On the construction of an artificial neural network for solving a system of equations Navier–Stokes in the case of incompressible fluid

Artificial Neural Networks to Predict Metabolic Syndrome without Invasive Methods in Adolescents

Artificial Neural Networks to Predict Metabolic Syndrome without Invasive Methods in Adolescents

Utilizing artificial neural network system to predict the residual valve after endoscopic posterior urethral valve ablation.

To build, train, and assess the artificial neural network (ANN) system in estimatingthe residual valve rate after endoscopic valve ablationand compare the data obtained with conventional analysis. In a retrospective cross-sectional study between June 2010 and December 2020, 144 children with a history of posteriorurethral valve (PUV) who underwent endoscopic valve ablation were enrolled in the study. MATLAB software was used to design and train the network in a feed-forward backpropagationerror adjustment scheme. Preoperative and postoperative data from 101 patients (70%) (training set) were utilized to assessthe impact and relative significance of the necessity for repeatedablation. The validated suitably trained ANN was used to predict repeated ablation in the next 33 patients (22.9%) (test set)whose preoperative data were serially input into the system. To assess system accuracy in forecasting the requirement for repeatablation, projected values were compared to actual outcomes. The likelihood of predicting the residual valve was calculatedusing a three-layered backpropagating deep ANN using preoperativeand postoperative information. Of 144 operated cases, 33 (22.9%) had residual valvesand needs to repeated ablation. The ANN accuracy, sensitivity, and specificity for predicting the residual valve were 90.75%,92.73%, and 73.19%, respectively. Younger age at surgery, hyperechogenicity of the renal parenchyma, presence of vesicoureteralreflux (VUR), and grade of reflux before surgery were among the most significant characteristics that affectedpostoperative outcome variables, the need for repeated ablation, and were given the highest relative weight by the ANN system. Conclusions: The ANN is an integrated data-gathering tool for analyzing and finding relationships among variables as a complex non-linear statistical model. The results indicate that ANN is a valuable tool for outcome prediction of the residualvalveafter endoscopic valve ablation in patients with PUV.