Number Of Layers Research Articles

Investigation of intra-fractionated range guided adaptive proton therapy (RGAPT): II. Range-shift compensated on-line treatment adaptation and verification

We previously proposed range-guided adaptive proton therapy (RGAPT) that uses mid-range treatment beams as probing beams and intra-fractionated range measurements for online adaptation. In this work, we demonstrated experimental verification and reported the dosimetric accuracy for RGAPT. A STEEV phantom was used for the experiments, and a 3 × 3 × 3 cm3 cube inside the phantom was assigned to be the treatment target. We simulated three online range shift scenarios: reference, overshoot, and undershoot, by placing upstream Lucite sheets, 4, 0, and 8 that corresponded to changes of 0, 6.8, and −6.8 mm, respectively, in water-equivalent path length. The reference treatment plan was to deliver single-field uniform target doses in pencil beam scanning mode and generated on the Eclipse treatment planning system. Different numbers of mid-range layers, including single, three, and five layers, were selected as probing beams to evaluate beam range (BR) measurement accuracy in positron emission tomography (PET). Online plans were modified to adapt to BR shifts and compensate for probing beam doses. In contrast, non-adaptive plans were also delivered and compared to adaptive plans by film measurements. The mid-range probing beams of three (5.55MU) and five layers (8.71MU) yielded accurate range shift measurements in 60 s of PET acquisition with uncertainty of 0.5 mm while the single-layer probing (1.65MU) was not sufficient for measurements. The adaptive plans achieved an average gamma (2%/2 mm) passing rate of 95%. In contrast, the non-adaptive plans only had an average passing rate of 69%. RGAPT planning and delivery are feasible and verified by the experiments. The probing beam delivery, range measurements, and adaptive planning and delivery added a small increase in treatment delivery workflow time but resulted in substantial dose improvement. The three-layer mid-range probing was most suitable considering the balance of high range measurement accuracy and the low number of probing beam layers.

Atomically thin-layered WS2 based resistive sensors for detection of CO and NO2 at room temperature

The number of layers present in a two-dimensional (2D) nanomaterial plays a critical role in applications that involve surface interaction, for example, gas sensing. This paper reports the synthesis of 2D WS2 nanoflakes using the facile liquid exfoliation technique. The nanoflakes were exfoliated using bath sonication (BS-WS2) and probe sonication (PS-WS2). The thickness of the BS-WS2 was found to range between 70 and 200 nm, and that of PS-WS2 varied from 0.6 to 80 nm, indicating the presence of single to few layers of WS2 when characterized using atomic force microscope. All the WS2 samples were thoroughly characterized using electron microscopes, x-ray diffractometer, Raman spectroscopy, UV–Visible spectroscopy, Fourier transform infrared spectroscope, and thermogravimetric analyser. Both the nanostructured samples were exposed to 2 ppm of NO2 at room temperature. Interestingly, BS-WS2 which comprises of a greater number of WS2 layers exhibited −14.2% response as against −3.4% response of PS-WS2, the atomically thin sample. The BS-WS2 sample was found to be highly selective towards NO2 but was slower (with incomplete recovery) as compared to PS-WS2. The PS-WS2 sample was observed to exhibit −11.9% to −27.4% response to 2–10 ppm of CO and −3.4%–35.2% response to 2–10 ppm of NO2 at room temperature, thereby exhibiting the potential to detect two gases simultaneously. These gases could be accurately predicted and quantified if the response times of the PS-WS2 sample were considered. The atomically thin WS2-based sensor exhibited a limit of detection of 131 and 81 ppb for CO and NO2, respectively.

Compressive performance of underwater concrete piers reinforced with prefabricated FRP shells

A novel underwater spliceable basalt fiber-reinforced polymer (BFRP) shell-confined concrete structure was proposed that rapidly reinforces underwater piers by lapping the FRP shell with adhesive underwater combined with nondispersive mortar (NDM) filling. Thirty-six prefabricated BFRP shells combined with nondispersive mortar reinforced concrete columns (PFRCs) were made for axial compression testing. The effects of the BFRP layer number, lap length and pouring environment on the specimen performance were compared, and different damage patterns and stressstrain relationship curves were obtained. The test results reveal that the PFRCs exhibited two failure modes: BFRP fracture failure in nonlapped segments and BFRP bond adhesive peeling failure in lap segments, and the specimens with more BFRP layers and shorter lap lengths were prone to BFRP bond adhesive peeling failure. From 2 to 4 layers, the ultimate stress and ultimate strain values increased by 1.18–1.91 times and 5.18–10.32 times, respectively, demonstrating increasing the number of BFRP layers is a more effective means to improve the structural performance. The ultimate strength of the underwater reinforced specimens reached more than 70 % of that of the corresponding specimens on land, which demonstrated the feasibility of using prefabricated FRP-molded shell reinforcing technology for submerged piers. Finally, the calculation model of an FRP-confined concrete column is evaluated, and the full stressstrain curves of the PFRCs are proposed.

Investigating layer-by-layer films of carbon nanotubes and nickel phthalocyanine towards diquat detection

The indiscriminate use of pesticides makes us susceptible to the toxicity of these chemical compounds, which may be present in high quantities in our food. It is crucial to develop inexpensive and rapid methods for determining these pesticides for government control or even for the general population. In this study, we investigated the fabrication of self-assembled LbL films using multi-walled carbon nanotubes (MWCNT) and nickel tetrasulphonated phthalocyanine (NiTsPc) as an electrochemical sensor for the herbicide Diquat (DQ). The Layer-by-Layer (LbL) assembly of the (MWCNT/NiTsPc) film was examined, along with its structural and morphological characteristics. The effect of the number of layers in DQ detection was evaluated by cyclic voltammetry, followed by the detection through differential pulse voltammetry. The achieved limit of detection was 9.62 × 10−7 mol L−1. A ~ 30% decrease in sensitivity was observed in the presence of Paraquat, a banned herbicide and electrochemical interferent due to the structural similarities, which is regularly neglected in the most published studies. The sensor was tested in real samples, demonstrating a recovery of 98.5% in organic apples.

Numerical Evaluation of the Influence of Using Carbon-Fiber-Reinforced Polymer Rebars as Shear Connectors for Cross-Laminated Timber–Concrete Panels

Carbon fiber-reinforced polymer (CFRP) sheets have been used to reinforce cross-laminated timber (CLT)–concrete systems in recent years. The existing studies have indicated that the use of CFRP rebars as shear connectors in CLT–concrete panels can improve the structural performance of these elements. However, the application and understanding of CFRP rebars as shear connectors still need to be improved, since comprehensive studies on the subject are not available. Therefore, this research aimed to evaluate the structural performance of CLT–concrete panels with CFRP rebars as shear connectors through finite element (FE) numerical simulation. A parametric study was conducted, varying the connector material, the number of CLT layers, the connector insertion angle, and the connector embedment length. According to the results, panels with CFRP connectors showed a higher maximum load, bending strength, and maximum bending moment than panels with steel connectors. The regression models revealed that the parameters analyzed explained between 80.2% and 99.9% of the variability in the mechanical properties under investigation. The high explanatory power (R2) of some regression models in this study underscores the robustness of the models. The number of CLT layers and the connector material were the most significant parameters for the panels’ maximum load, displacement at the maximum load, ductility, bending strength, and maximum bending moment. The number of CLT layers and the connector insertion angle were the most significant parameters for the panels’ effective bending stiffness. This research highlights the importance of studies on CLT–concrete composites and the need to develop equations to estimate their behavior accurately. Moreover, numerical simulations have proven very valuable, providing results comparable to laboratory results.

Molecular Dynamics-Based Study of Graphene/Asphalt Mechanism of Interaction

This study employed molecular dynamics simulation to investigate the mechanism of action of graphene-modified asphalt. A series of molecular models of graphene-modified asphalt were constructed and validated using thermodynamic parameters. The impact of the graphene (PGR) size and number of layers on its interaction with asphalt components were examined, and the self-healing process and mechanism of action of PGR-modified asphalt were analyzed. The results demonstrated that the size and number of layers of PGR significantly influenced its interaction with asphalt components, with polar components demonstrating a stronger affinity for PGR. When the size and number of layers of PGR were held constant, the interfacial binding energy between it and ACR-modified asphalt was the highest, followed by SBS-modified asphalt, and 70# matrix asphalt exhibited the lowest interfacial binding strength. This interfacial binding strength is primarily attributed to intermolecular van der Waals interactions. Furthermore, the incorporation of multi-layer PGR can markedly enhance the mechanical properties of matrix asphalt, whereas small-sized PGR is more efficacious in improving the low-temperature performance of polymer-modified asphalt. PGR can act as a bridge between asphalt molecules through rapid heat transfer and π-π stacking with aromatic ring-containing substances, which markedly increases the free diffusion ability of asphalt molecules, shortens the healing time of asphalt, and enhances the collective self-healing performance of asphalt. This study provides an essential theoretical basis for understanding the mechanism and application of PGR in asphalt modification.

Engineering optoelectronic properties of [formula omitted] superlattices using first principles method

Recent advancements in thin-film growth techniques have opened doors for engineering innovative heterostructures with ultra-thin layers. These artificial superlattices hold promise for novel optoelectronic devices. However, a complete understanding of their properties is crucial for optimal design. In this study, we employ the full-potential linearized augmented plane wave (FP−LAPW) approach based on density functional theory (DFT) to engineer the optoelectronic properties of (HgSe)n/(ZnTe)n superlattices by controlling the number of layers (n) in SLs; The exchange-correlation potential was calculated by the generalized gradient GGA approximation and the optoelectronics properties has adjusted by the Tran-Blaha modified Becke-Johnson (TB−mBJ) correction of GGA approximation. Our findings shed light on the significant influence of stacking periodicity on the optoelectronic properties of HgTe/ZnTe SLs. We demonstrate that manipulating layer count is a viable strategy for engineering their optoelectronic characteristics. Additionally, the predicted near-infrared absorption makes these SLs promising candidates for near-infrared detector applications.

Strategic Geosteering Workflow with Uncertainty Quantification and Deep Learning: Initial Test on the Goliat Field Data

Continuous integration of real-time logging-while-drilling data into a subsurface model with relevant geological uncertainties enables strategic geosteering: a field-level optimization of the well-placement strategy. Model errors arising from oversimplified conceptual geological models and imperfect simulation of measurements result in unreliable subsurface-model updates. The model errors are particularly pronounced when synthetic measurements are approximated with a fast but imperfect model, such as a deep neural network (DNN).#xD;We present a practical data-assimilation workflow consisting of offline and online phases. The offline phase involves DNN training and building an uncertain prior near-well geo-model. The online phase utilizes the flexible iterative ensemble smoother (FlexIES) to perform real-time assimilation of extra-deep electromagnetic data while accounting for the model errors in the approximate DNN model. We demonstrate the proposed workflow on historic well-log data from the Goliat Field (Barents Sea). #xD;The median of our probabilistic estimation is on par with proprietary inversion, regardless of the number of layers in the chosen prior or the approximate DNN model. By estimating model errors, FlexIES automatically quantifies the uncertainty in the boundaries and resistivities of layers, which is not standard in proprietary inversion. #xD;This capability allows us to capture uncertainties more efficiently, thus providing input for future quantitative decision support methods. We demonstrate the potential of quantitive decision support by visually estimating the ahead-of-bit risk of reservoir exit that has occurred during the considered operation.

Optimizing feedforward neural networks using a modified weighted mean of vectors: Case study chemical datasets

This paper proposes a modified version of the weighted mean of vectors algorithm (mINFO), which combines the strengths of the INFO algorithm with the Enhanced Solution Quality Operator (ESQ). The ESQ boosts the quality of the solutions by avoiding optimal local values, verifying that each solution moves towards a better position, and increasing the convergence speed. Furthermore, we employ the mINFO algorithm to optimize the connection weights and biases of feedforward neural networks (FNNs) to improve their accuracy. The efficacy of FNNs for classification tasks is mainly dependent on hyperparameter tuning, such as the number of layers and nodes. The mINFO was evaluated using the IEEE Congress on Evolutionary Computation held in 2020 (CEC’2020) for optimization tests, and ten chemical data sets were applied to validate the performance of the FNNs classifier. The proposed algorithm’s results have been evaluated with those of other well-known optimization methods, including Runge Kutta optimizer’s (RUN), particle swarm optimization (PSO), grey wolf optimization (GWO), Harris hawks optimization (HHO), whale optimization algorithm (WOA), slime mould algorithm (SMA) and the standard weighted mean of vectors (INFO). In addition, some improved metaheuristic algorithms. The experimental results indicate that the proposed mINFO algorithm can improve the convergence speed and generate effective search results without increasing computational costs. In addition, it has improved the FNN’s classification efficiency.

Fully Printed Multilayer Ceramic Capacitors Based on High-k Perovskite Nanosheets.

Printing technology enables the integration of chemically exfoliated perovskite nanosheets into high-performance microcapacitors. Theoretically, the capacitance value can be further enhanced by designing and constructing multilayer structures without increasing the device size. Yet, issues such as interlayer penetration in multilayer heterojunctions constructed using inkjet printing technology further limit the realization of this potential. Herein, a series of multilayer configurations, including Ag/(Ca2NaNb4O13/Ag)n and graphene/(Ca2NaNb4O13/graphene)n (n = 1-3), are successfully inkjet-printed onto diverse rigid and flexible substrates through optimized ink formulations, inkjet printing parameters, thermal treatment conditions, and rational multilayer structural design using high-k perovskite nanosheets, graphene nanosheets and silver. The dielectric performance is optimized by fine-tuning the number of dielectric layers and modifying the electrode/dielectric interface. As a result, the graphene/(Ca2NaNb4O13/graphene)3 multilayer ceramic capacitors exhibit a remarkable capacitance density of 346±12nFcm-2 and a high dielectric constant of 193±18. Additionally, these devices demonstrate moderate insulation properties, flexibility, thermal stability, and chemical sensitivity. This work shed light on the potential of multilayer structural design in additive manufacturing of high-performance 2D material-based ceramic capacitors.

Highly-efficient low-cost polarization-sensitive organic photodetectors based on laminated self-assembly planar and bulk heterojunctions

To overcome the severe problems arising from the insufficient light absorption of ultrathin self-assembly active layers and the high cost use of atomic force deposition (ALD)-grown low-leakage-current transport layers, we successfully developed a low-cost, simple and facile strategy of floating-film transfer and multilayer lamination (FFTML) for constructing highly-efficient ALD-free broadband polarization-sensitive organic photodetectors (OPDs) with the two commonly used structures of donor/acceptor planar heterojunction (PHJ) and donor:acceptor multilayer bulk heterojunction (BHJ). It was found that the PHJ-based polarization-sensitive OPD by FFTML possesses a low dark current due to the high carrier injection barrier, indicating it is more suitable to be applied in low polarized light detection scenarios. In contrast, the BHJ-based device by FFTML has a higher spectral responsivity in the whole wavelength due to more photo-excitons transferred to the donor:acceptor interface and dissociated into photoexcited carrirers. Furthermore, the film thickness, which is tuned by increasing lamination number of BHJ layers, has a big effect on the polarization-sensitive photodetection performance. The polarization-sensitive 4-BHJ OPD by FFTML finally achieved a high specific detectivity of 8.33 × 1010 Jones, which was much higher than 2.72 × 1010 Jones for the 2-BHJ device at 0 V. This work demonstrates that layer-by-layer lamination of self-assembly films can effectively improve the polarized-light detection performance, contributing significantly to the rapid development of the field.

The effect of classical optimizers and Ansatz depth on QAOA performance in noisy devices

The Quantum Approximate Optimization Algorithm (QAOA) is a variational quantum algorithm for Near-term Intermediate-Scale Quantum computers (NISQ) providing approximate solutions for combinatorial optimization problems. The QAOA utilizes a quantum-classical loop, consisting of a quantum ansatz and a classical optimizer, to minimize some cost function, computed on the quantum device. This paper presents an investigation into the impact of realistic noise on the classical optimizer and the determination of optimal circuit depth for the Quantum Approximate Optimization Algorithm (QAOA) in the presence of noise. We find that, while there is no significant difference in the performance of classical optimizers in a state vector simulation, the Adam and AMSGrad optimizers perform best in the presence of shot noise. Under the conditions of real noise, the SPSA optimizer, along with ADAM and AMSGrad, emerge as the top performers. The study also reveals that the quality of solutions to some 5 qubit minimum vertex cover problems increases for up to around six layers in the QAOA circuit, after which it begins to decline. This analysis shows that increasing the number of layers in the QAOA in an attempt to increase accuracy may not work well in a noisy device.

Numerical and experimental study of attenuation inspection during FDM with ultrasonic leaky waveguide

ABSTRACT This work proposes an ultrasonic leaky waveguide (ULWG) for in-situ structural integrity inspection during fused deposition modeling (FDM). Since the ultrasonic wave propagation speed (m·s−1) is much higher than the deposition speed (mm·s−1), a quasi-static finite element model is first established to simulate the wave propagation during this dynamic process. The amplitude and energy attenuation coefficients (α and ζ) are chosen as the inspection features. Processing parameters including deposition layer thickness and number, inspection frequency, plate thickness and interface impedance are explored. For the first layer printing, a thinner thickness, higher frequency and proper interface impedance will enable a better inspection with high sensitivity. For multilayer printing, the multimodal dispersion curves change continuously and α oscillation is observed due to waveform aliasing that ζ is more reliable. Printing failures such as detachment can be predicted as well from the model established. The ULWG is then integrated into a commercial FDM machine to capture the attenuation coefficients of an “S” shape sample. The poor adhesion, local and global detachments, path planning impact can all be revealed by ζ experimentally. Moreover, being consistent to the simulation, the fluctuation of α in multilayer printing increases, which prevents a further printing evaluation.

Numerical and experimental study of attenuation inspection during FDM with ultrasonic leaky waveguide

ABSTRACT This work proposes an ultrasonic leaky waveguide (ULWG) for in-situ structural integrity inspection during fused deposition modeling (FDM). Since the ultrasonic wave propagation speed (m·s−1) is much higher than the deposition speed (mm·s−1), a quasi-static finite element model is first established to simulate the wave propagation during this dynamic process. The amplitude and energy attenuation coefficients (α and ζ) are chosen as the inspection features. Processing parameters including deposition layer thickness and number, inspection frequency, plate thickness and interface impedance are explored. For the first layer printing, a thinner thickness, higher frequency and proper interface impedance will enable a better inspection with high sensitivity. For multilayer printing, the multimodal dispersion curves change continuously and α oscillation is observed due to waveform aliasing that ζ is more reliable. Printing failures such as detachment can be predicted as well from the model established. The ULWG is then integrated into a commercial FDM machine to capture the attenuation coefficients of an “S” shape sample. The poor adhesion, local and global detachments, path planning impact can all be revealed by ζ experimentally. Moreover, being consistent to the simulation, the fluctuation of α in multilayer printing increases, which prevents a further printing evaluation.

Probing electronic and dielectric properties of ultrathin Ga2O3/Al2O3 atomic layer stacks made with in vacuo atomic layer deposition

Ultrathin (1–4 nm) films of wide-bandgap semiconductors are important to many applications in microelectronics, and the film properties can be sensitively affected by defects especially at the substrate/film interface. Motivated by this, an in vacuo atomic layer deposition (ALD) was developed for the synthesis of ultrathin films of Ga2O3/Al2O3 atomic layer stacks (ALSs) on Al electrodes. It is found that the Ga2O3/Al2O3 ALS can form an interface with the Al electrode with negligible interfacial defects under the optimal ALD condition whether the starting atomic layer is Ga2O3 or Al2O3. Such an interface is the key to achieving an optimal and tunable electronic structure and dielectric properties in Ga2O3/Al2O3 ALS ultrathin films. In situ scanning tunneling spectroscopy confirms that the electronic structure of Ga2O3/Al2O3 ALS can have tunable bandgaps (Eg) between ∼2.0 eV for 100% Ga2O3 and ∼3.4 eV for 100% Al2O3. With variable ratios of Ga:Al, the measured Eg exhibits significant non-linearity, agreeing with the density functional theory simulation, and tunable carrier concentration. Furthermore, the dielectric constant ε of ultrathin Ga2O3/Al2O3 ALS capacitors is tunable through the variation in the ratio of the constituent Ga2O3 and Al2O3 atomic layer numbers from 9.83 for 100% Ga2O3 to 8.28 for 100% Al2O3. The high ɛ leads to excellent effective oxide thickness ∼1.7–2.1 nm for the ultrathin Ga2O3/Al2O3 ALS, which is comparable to that of high-K dielectric materials.

Structural and ferroelectric transition in few-layer HfO2 films from first principles

Abstract The discovery of ferroelectricity in HfO2-based materials with high dielectric constant has inspired tremendous research interest for next-generation electronic devices. Importantly, films structure and strain are key factors in the exploration of ferroelectricity in the fluorite-type oxide HfO2 films. Here we have investigated the structures and strain-induced ferroelectric transition in the different phases of few-layer HfO2 films (layer number N=1-5). It is found that HfO2 films for all phases are more stable with increasing films thickness. Among them, the Pmn2 1 (110)-oriented film is the most stable, and the films of N=4, 5 occur a P2 1 ferroelectric transition under tensile strain, resulting in polarization about 11.8 μC/cm2 along in-plane a-axis. The ferroelectric transition is caused by the strain, which induces the displacement of Hf and O atoms on the surface to non-centrosymmetric positions away from the original paraelectric positions, accompanied by the change of surface Hf-O bond lengths. More importantly, three new stable HfO2 2D structures are discovered, together with computed electronic structures, mechanical, and dielectric properties analyses. This work provides guidance for theoretical and experimental study of the new structures and strain-tuned ferroelectricity in freestanding HfO2 films.

Swelling characteristics of montmorillonite mineral particles in Gaomiaozi bentonite

Highly compacted bentonite is an ideal back-filling (buffer) material for the deep geological disposal of high-level radioactive waste. Montmorillonite is the main constituent mineral of it. The nanoscopic swelling characteristic of montmorillonite mineral particles greatly influences the macroscopic buffering performance of highly compacted bentonite blocks. Firstly, the montmorillonite suspension liquid was exfoliated by freeze-thaw cycles, the appropriate mineral particles were selected by atomic force microscope (AFM) for making colloidal probes. Then, the swelling pressure between mineral particles was measured by AFM, investigating the effects of interlayer distance, layer number, temperature, and interlayer cation type on the nanoscale swelling pressure. The experimental results showed that the swelling pressure of montmorillonite mineral particles peaks at about 0.7–0.8 nm interlayer distance. The swelling pressure of mineral particles with more layers is smaller. The higher the ambient temperature, the greater the swelling pressure. The swelling pressure of Ca-montmorillonite is significantly larger than that of Na-montmorillonite. The Laird model has a relatively good applicability with the interlayer distance of 0.6–1.2 nm; the DLVO theory has a good applicability with the interlayer distance greater than 1.5 nm. Finally, a predictive model was developed for the swelling pressure of nanoscopic mineral particles considering the effect of layer number. The research results provide data support and theoretical support for swelling pressure generation and cross-scale transmission mechanism of highly compacted bentonite hydration.

Evaluation of artificial neurocomputing algorithms and their metacognitive robustness in predictive modeling of fuel consumption rates during tillage

Artificial intelligence (AI) requires complex neurocomputing algorithms (NCAs) for robustness in predictive modeling. Since 1990, a plethora of NCAs have evolved, challenging selection of the most intelligent and error-tolerant model for predicting fuel consumption rate (ØFuel) in tillage. This study adopted 12 recent NCAs in modeling 72 neurocomputing architectures and evaluated their metacognitive robustness in predicting ØFuel using 80 tillage datasets. These included Levenberg-Marquardt (trainlm), Quasi-newton (trainbfg), Powell-Beale conjugate gradient (traincgb), Scaled conjugate gradient (trainscg), Fletcher-reeves conjugate gradient (traincgf), Polak-ribiére conjugate gradient (traincgp), Onestep secant (trainoss), Bayesian regularization (trainbr), Resilient backpropagation (trainrp), Gradient descent (traingd), Learning rate gradient descent (traingdx) and Gradient descent momentum (traingdm). Tillage variables were sequentially subjected to NCA architectures learning on logsig, Purelin, and tansig neuro-activation functions, targeting 1000 epochs. Neuro-cognitive performance was evaluated through broad multi-criteria of regressed neuron input–output functions using heuristic and accuracy metrics, i.e., training time (t), number of hidden layers (NL), and neurons (Nn), optimal epoch (Eopt), Coefficient of correlation (R) and determination (R2), Mean square error (MSE), Root mean square error (RMSE), Mean absolute error (MAE), Mean absolute percentage error (MAPE), Sum square error (SSE), Prediction scatter (Tscatter), Coefficient of variation (CV) and Prediction accuracy (PA). Prediction reliability and robustness were evaluated using a20-index (a20), Willmott’s index of agreement (IOA), Index of scatter (IOS), Variance accounted for (VAF), Performance index (PI), Total score rank (ξtotal), Sensitivity index (SAij) and Anderson-Darling (AD) test. Results indicated that varying NL and Nn did not correspond with t, Eopt and PA but only occurred in single-layered trainbr and double-layered trainlm. Although PA increased at reduced convergence error limits, relationships among Eopt, t, and PA were unclear. Single-layered trainbr (14-20-1) outperformed all NCAs with R (0.9993), R2 (0.9986), MSE (6.34e-06), RMSE (0.0025), MAE (0.003), MAPE (1.4e-06), SSE (8e-04), R2All (0.9998) Tvalue (1.0), CV (3.232 %) and PA (99.81 %). However, trainlm (14-20-10-1) most accurately modeled double-layered architectures i.e. R (0.9932), R2 (0.9864), MSE (7.4e-05), RMSE (0.009), MAE (0.004), MAPE (0.003), SSE (0.003), R2All (0.9929), Tvalue (0.9998), CV (3.212 %) and PA (97.8 %). Nonetheless, trainbr (14-20-1) achieved the best reliability metrics of a20 (100 %), VAF (99.98 %), IOA (1.00), PI (1.9971), IOS (0.0115), ξtotal (284) and AD-test (2.6852). Sensitivity analysis revealed that tractive force had the most significant influence on ØFuel while tire inflation pressure had the least. Learning on transig neuro-activation function in single-layered neural network, metacognitive robustness of trainbr is superiorly merited for predicting ØFuel during tillage.

AMIKOMNET: Novel Structure for a Deep Learning Model to Enhance COVID-19 Classification Task Performance

Since early 2020, coronavirus has spread extensively throughout the globe. It was first detected in Wuhan, a province in China. Many researchers have proposed various models to solve problems related to COVID-19 detection. As traditional medical approaches take a lot of time to detect the virus and require specific laboratory tests, the adoption of artificial intelligence (AI), including machine learning, might play an important role in handling the problem. A great deal of research has seen the adoption of AI succeed in the early detection of COVID-19 using X-ray images. Unfortunately, the majority of deep learning adoption for COVID-19 detection has the shortcomings of high error detection and high computation costs. In this study, we employed a hybrid model using an auto-encoder (AE) and a convolutional neural network (CNN) (named AMIKOMNET) with a small number of layers and parameters. We implemented an ensemble learning mechanism in the AMIKOMNET model using Adaboost with the aim of reducing error detection in COVID-19 classification tasks. The experimental results for the binary class show that our model achieved high effectiveness, with 96.90% accuracy, 95.06% recall, 94.67% F1-score, and 96.03% precision. The experimental result for the multiclass achieved 95.13% accuracy, 94.93% recall, 95.75% F1-score, and 96.19% precision. The adoption of Adaboost in AMIKOMNET for the binary class increased the effectiveness of the model to 98.45% accuracy, 96.16% recall, 95.70% F1-score, and 96.87% precision. The adoption of Adaboost in AMIKOMNET in the multiclass classification task also saw an increase in performance, with an accuracy of 96.65%, a recall of 94.93%, an F1-score of 95.76%, and a precision of 96.19%. The implementation of AE to handle image feature extraction combined with a CNN used to handle dimensional image feature reduction achieved outstanding performance when compared to previous work using a deep learning platform. Exploiting Adaboost also increased the effectiveness of the AMIKOMNET model in detecting COVID-19.

AMIKOMNET: Novel Structure for a Deep Learning Model to Enhance COVID-19 Classification Task Performance

Since early 2020, coronavirus has spread extensively throughout the globe. It was first detected in Wuhan, a province in China. Many researchers have proposed various models to solve problems related to COVID-19 detection. As traditional medical approaches take a lot of time to detect the virus and require specific laboratory tests, the adoption of artificial intelligence (AI), including machine learning, might play an important role in handling the problem. A great deal of research has seen the adoption of AI succeed in the early detection of COVID-19 using X-ray images. Unfortunately, the majority of deep learning adoption for COVID-19 detection has the shortcomings of high error detection and high computation costs. In this study, we employed a hybrid model using an auto-encoder (AE) and a convolutional neural network (CNN) (named AMIKOMNET) with a small number of layers and parameters. We implemented an ensemble learning mechanism in the AMIKOMNET model using Adaboost with the aim of reducing error detection in COVID-19 classification tasks. The experimental results for the binary class show that our model achieved high effectiveness, with 96.90% accuracy, 95.06% recall, 94.67% F1-score, and 96.03% precision. The experimental result for the multiclass achieved 95.13% accuracy, 94.93% recall, 95.75% F1-score, and 96.19% precision. The adoption of Adaboost in AMIKOMNET for the binary class increased the effectiveness of the model to 98.45% accuracy, 96.16% recall, 95.70% F1-score, and 96.87% precision. The adoption of Adaboost in AMIKOMNET in the multiclass classification task also saw an increase in performance, with an accuracy of 96.65%, a recall of 94.93%, an F1-score of 95.76%, and a precision of 96.19%. The implementation of AE to handle image feature extraction combined with a CNN used to handle dimensional image feature reduction achieved outstanding performance when compared to previous work using a deep learning platform. Exploiting Adaboost also increased the effectiveness of the AMIKOMNET model in detecting COVID-19.