Number Of Intermediate Layers Research Articles

The response of household debt to COVID-19 using a neural networks VAR in OECD.

This paper investigates responses of household debt to COVID-19-related data like confirmed cases and confirmed deaths within a neural networks panel VAR for OECD countries. Our model also includes a plethora of non-pharmaceutical and pharmaceutical interventions. We opt for a global neural networks panel VAR (GVAR) methodology that nests all OECD countries in the sample. Because linear factor models are unable to capture the variability in our data set, the use of an artificial neural network (ANN) method permits to capture this variability. The number of factors, as well as the number of intermediate layers, is determined using the marginal likelihood criterion and we estimate the GVAR with MCMC techniques. We also report δ-values that capture the dominance of each individual country in the network. In terms of dominant countries, the UK, the USA, and Japan dominate interconnections within the network, but also countries like Belgium, Netherlands, and Brazil. Results reveal that household debt positively responds to COVID-19 infections and deaths. Lockdown measures such as stay-at-home advice, and closing schools, all have a positive impact on household debt, though they are of transitory nature. However, vaccinations and testing appear to negatively affect household debt.

A Hybrid Methodology Based on Machine Learning for a Supply Chain Optimization Problem

This paper presents an advanced methodology that integrates a machine learning methodology into an optimization process. The framework of an interactive machine learning algorithm was developed to meet the challenges in solving large-scale optimization problems. An artificial neural network (ANN) is used with the knowledge gained from solving previous problems with different scenarios to define a good starting point for a solution searching process. By using an initial solution, known as “warm start”, the search space can be reduced to get more opportunity to find an optimal solution. The applicability of the proposed method was evaluated by using it to determine the optimal facility locations for a biomass supply chain problem using a real case study from Central Vietnam. The supply chain planning model is based on an optimization model, where the goal is to maximize the benefits from meeting the electricity demand minus the total cost from facility cost, penalty cost from lost demand, and operational costs form the supply chain. The structure of the ANN, the number of intermediate layers and the number of processing nodes, was determined by comparing the accuracy from different configurations. The ANN with two intermediate layers possesses the best performance from the training and testing datasets. The proposed model succeeded in predicting the facility location with more than 98% prediction accuracy. The results from our framework provide optimal solutions while saving runtime.

Machine Learning Model for a Dynamic Path Planning Problem

Due to an advancement in Industry 4.0 technology, various autonomous systems have been developed in order to increase the operational efficiency. This paper considers an application of Industry 4.0 technology to an autonomous transportation operation. The paper focuses on applying a machine learning technique to a dynamic path planning problem where real-time randomized obstacle locations are considered. The routes or the solutions from the dynamic path planning problem are determined by an A-star algorithm, which are then used to build machine learning models based on an artificial neural network. The models were developed to discover the relationship between the input and output of the dynamic path planning problem. The structure of the network which is defined by the number of intermediate layers and the number of nodes is provided, where the overall accuracy is used to evaluate the setting efficiency. The proposed methodology was tested with a problem that consists of 7 types of paths, and the number of randomized obstacles fluctuated from 1 to 8. The paths were generated based on a layout of a consumer product warehouse. The proposed model succeeded in predicting the robot paths with 98.5% prediction accuracy.

Ultrasound Imaging through Layers with Unknown Parameters

Abstract—A method is presented for generating ultrasound images in imaging devices with phased arrays, based on decomposition of the recorded spatiotemporal signal into a spectrum of plane pulsed waves. A case is considered when a number of intermediate layers with unknown thicknesses and sound speeds are located between the ultrasound array and the visualization region. The method is based on measuring the delays of the spectral components of plane waves transmitted through layers and reflected from the boundary of the visualization region, as well as on compensating for these delays when summing the spectral components scattered by inhomogeneities in the visualization region.

A Neuronal Morphology Classification Approach Based on Locally Cumulative Connected Deep Neural Networks

Neurons are the basic building and computational units of the nervous system, and have complex and diverse spatial geometric structures. By solving the neuronal classification problem, we can further understand the characteristics of neurons and the process of information transmission. This paper presents a neuronal morphology classification approach based on locally cumulative connected deep neural networks, where 43 geometric features were extracted from two different neuron datasets and applied to classify types of neurons. Then, the effects of different parameters of deep learning networks on the performance of neuron classification were analyzed including mini-batch size, number of intermediate layers, and number of building blocks. The accuracy of the approach was also compared with that of the other mainstream machine learning approaches. The experimental results showed that the proposed approach is effective for solving complex neuronal morphology classification problems.

CATSMLP: Toward a Robust and Interpretable Multilayer Perceptron With Sigmoid Activation Functions

Enhancing the robustness and interpretability of a multilayer perceptron (MLP) with a sigmoid activation function is a challenging topic. As a particular MLP, additive TS-type MLP (ATSMLP) can be interpreted based on single-stage fuzzy IF-THEN rules, but its robustness will be degraded with the increase in the number of intermediate layers. This paper presents a new MLP model called cascaded ATSMLP (CATSMLP), where the ATSMLPs are organized in a cascaded way. The proposed CATSMLP is a universal approximator and is also proven to be functionally equivalent to a fuzzy inference system based on syllogistic fuzzy reasoning. Therefore, the CATSMLP may be interpreted based on syllogistic fuzzy reasoning in a theoretical sense. Meanwhile, due to the fact that syllogistic fuzzy reasoning has distinctive advantage over single-stage IF-THEN fuzzy reasoning in robustness, this paper proves in an indirect way that the CATSMLP is more robust than the ATSMLP in an upper-bound sense. Several experiments were conducted to confirm such a claim.

Optimization of the parameters of an acoustic delay line by applying a rear acoustic load to a piezoelectric transducer

The operation of a microwave piezoelectric transducer whose piezoelectric element is connected through a number of intermediate layers with two acoustic loads is investigated. One of the loads is the medium in which the transducer excites elastic waves, and the other (the rear load) serves for the formation of the transducer’s characteristics. On the basis of the previous numerical analysis of the transfer coefficients for the direct and inverse transformations, the frequency dependences of the transformation efficiency and the standing-wave ratios in the transmission line (the direct transformation) and in the sound duct (the inverse transformation) are calculated. The theoretical curves that illustrate the effect of the rear load on the characteristics of a specific multilayer transducer operating in the decimetric wave range with a matching LC circuit are discussed.

Transfer constant of a multilayer acoustic transducer at the direct and inverse transformation

The algorithms for calculating the direct and inverse transfer constants of an acoustic transducer with an arbitrary number of intermediate layers between the piezoelectric layer, the acoustic duct, and the rear acoustic load are described. The results of a numerical analysis are presented and discussed. As an illustration, a 100-MHz transducer formed by a (Y+36°)-cut LiNbO3 plate fixed on a fused-quartz acoustic duct with the help of five metal layers is considered. The other side of the plate carries two metal layers and a rear load. The phase-frequency characteristics and the transformation loss as a function of frequency are analyzed for the cases of direct and inverse transformation under the assumption that the signal is supplied and retrieved by a two-wire line.

Graded coatings by gradient temperature densification

Functionally gradient materials (FGM) were prepared using layers of ZrO2 –3 mol% Y2O3 ceramic and NiCrAlY powders. A fine-grained zirconia powder was chosen to lower the ceramic sintering temperature and achieve simultaneous metal and ceramic densification. Consolidation of FGM's was achieved by a short time field-assisted sintering technique. Sintering was performed either at a constant temperature or in a temperature gradient by using punches made of different materials (i.e., one graphite and one tungsten). A temperature gradient of at least 100 °C was required with a low value of 1200 °C at the metal end and exceeding 1300 °C at the ceramic end. Increasing the number of intermediate layers alleviates some of the cracks formed during sintering due to different coefficients of thermal expansion.

Characterization of Thin-Film Electroluminescent Devices with Multiple Ta2O5 Interlayers Incorporated into SrS:Pr,Ce Phosphor

A thin-film electroluminescent device (ELD) has been fabricated by incorporating multiple Ta2O5 interlayers into the SrS:Pr,Ce phosphor layer using rf-magnetron sputtering. X-ray measurement results confirm the enhanced crystallinity of the phosphor layer and can be explained by the extra thermal-cycle annealing due to the different deposition temperatures for Ta2O5 (100° C) and SrS (500° C). The inclusion of thin barrier films is expected to redistribute the internal field and introduce additional interface states in the ELD structure. Based on the results of Sawyer-Tower circuit measurements, it is found that the transferred charge density in the active layer can increase with the number of intermediate layers. This contributes to an increase in brightness and the reduction of the threshold voltage. Moreover, a stacked structure for SrS:Pr and SrS:Ce phosphor layers with window effects is proposed to alleviate the absorption problem in the SrS:Pr,Ce multi-barrier ELD structure. Details of the mechanism and the improvement in the chromaticity will also be described.

Finite element comparative study of elastic stresses in single, double layer and multilayered coated systems

In this paper, a comparative study of elastic stresses in various types of coated surfaces including surfaces with single, double and multilayered coatings is described. The stresses have been computed for several coating thicknesses, one outer layer coating/substrate modulus ratio, and a wide range of friction coefficients. Stresses have been computed for normal load only and for combined normal and tangential loads. The main results of this study indicate that, for any given outer layer/substrate modulus ratio, the maximum tensile stress at the surfaces of all the coating systems considered increases with friction coefficient and decreases with coating thickness. Comparison of single layer, double layer and multilayered coating systems shows that, for a given total thickness and outer layer/substrate modulus ratio, the tensile stresses at the layer/substrate interface are progressively reduced as the number of intermediate layers in a multilayer coating is increased.

Influence of the Unit Number of Intermediate Layers and Networks on Learning Ability

Learning ability of networks in the recognition of vocabulary, etc. using neural networks is notably influenced by the unit number of intermediate layers. In our current study, we have investigated the effects when the unit number of intermediate layers is changed during the course of learning when back-propagation is applied, thereby discussing eventual influence on the learning ability of networks.