Minimum Number Of Layers Research Articles

Gaussian Aquila optimizer based dual convolutional neural networks for identification and grading of osteoarthritis using knee joint images

Degenerative musculoskeletal disease known as Osteoarthritis (OA) causes serious pain and abnormalities for humans and on detecting at an early stage, timely treatment shall be initiated to the patients at the earliest to overcome this pain. In this research study, X-ray images are captured from the humans and the proposed Gaussian Aquila Optimizer based Dual Convolutional Neural Networks is employed for detecting and classifying the osteoarthritis patients. The new Gaussian Aquila Optimizer (GAO) is devised to include Gaussian mutation at the exploitation stage of Aquila optimizer, which results in attaining the best global optimal value. Novel Dual Convolutional Neural Network (DCNN) is devised to balance the convolutional layers in each convolutional model and the weight and bias parameters of the new DCNN model are optimized using the developed GAO. The novelty of the proposed work lies in evolving a new optimizer, Gaussian Aquila Optimizer for parameter optimization of the devised DCNN model and the new DCNN model is structured to minimize the computational burden incurred in spite of it possessing dual layers but with minimal number of layers. The knee dataset comprises of total 2283 knee images, out of which 1267 are normal knee images and 1016 are the osteoarthritis images with an image of 512 × 512-pixel width and height respectively. The proposed novel GAO-DCNN system attains the classification results of 98.25% of sensitivity, 98.93% of specificity and 98.77% of classification accuracy for abnormal knee case–knee joint images. Experimental simulation results carried out confirms the superiority of the developed hybrid GAO-DCNN over the existing deep learning neural models form previous literature studies.

Knowledge and Practice of Face Mask Use and Handling during COVID-19 Pandemic among Dental Students and Intern at B.P. Koirala Institute of Health Sciences

Background: The outbreak of Corona Virus Diseases (COVID-19) has been deemed a global public health emergency by the World Health Organization (WHO). Face mask is one of the prevalent used methods to reduce the spread of infection. Due to their widespread use, it is crucial to use these masks properly because improper use or disposal could potentially speed up the pace of transmission. The objective was to assess the knowledge and practice of face mask use and handing in undergraduate dental students. Methods: A web based cross sectional study was conducted among BDS third, fourth, fifth year and interns at College of Dental Surgery, B. P. Koirala Institute of Health Sciences between May-June 2021. A structured proforma was prepared and modified based on relevant literature and expert’s advice. The Google Forms with questionnaires were created using docs.google.com/forms, and the link was shared with the signed-up participants via messaging apps Viber and WhatsApp. Statistical Package for the Social Sciences software, version 21 was used to analyse descriptive statistics such as frequency and percentage. Results: A total of 144 participants responded to the questionnaire. Eighty-five (59.03%) participants responded that minimum numberof layers in face mask that work better to help stop the spread of COVID-19 outside health care setting is three. Sixty-seven (46.53%) participants responded that they touch front surface of the face mask while wearing it. Conclusion: Campaigns to raise awareness about the proper use of face masks during this pandemic will be useful in educating dentistry students and interns.

Novel deep learning models with novel integrated activation functions for autism screening: AutiNet and MinAutiNet

Deep Learning (DL) particularly CNN has the potential to outplay conventional Machine Learning (ML) procedures in the realm of computer vision. With the advent of DL, profound research on autism utilizing fMRI data has declared advancements in autism screening. Yet, DL models are inadequate for heterogeneous datasets encompassing complex fMRI data. To resolve these problems, two cutting-edge AutiNet and MinAutiNet models are proposed with a minimum number of layers and neurons in classifying the autistic subjects. Both AutiNet and MinAutiNet models are the two refined forms of the traditional LeNet-5 model. AutiNet model includes two convolutional layers, two average pooling layers, a flatten layer, and one fully connected layer. One convolutional layer, one average pooling layer, one flatten layer, and one fully connected layer make up the MinAutiNet model. Pre-processing is a pre-requisite phase in the AutiNet and MinAutiNet models where denoising of raw fMRI data is done with slice time correction, realignment, min–max normalization and one hot encoding to obtain numerical values. As inspired by the promising results of non-handcrafted feature extraction, the proposed DL models have been used for feature extraction. Two novel integrated activation functions namely Li-ReLU and S-RReLU are developed and incorporated into the proposed DL models towards improving the performance of autism screening. Both AutiNet and MinAutiNet models surpass the standard accuracy in classifying autistic subjects with minimum number of layers and neurons. Experimental results reveal that the proposed AutiNet attains the maximum accuracy of 77.78% with Caltech and KKI datasets and the proposed MinAutiNet achieves the maximum accuracy of 88.89% with SBL dataset. The proposed models are compared with other state-of-the-art works to ensure its superiority.

Asymmetric polyelectrolyte multilayer nanofiltration membranes: Structural characterisation via transport phenomena

One promising application of polyelectrolyte multilayers (PEMs) is their use as selective layers in nanofiltration (NF). Especially encouraging are the simplicity of fabrication, controlled layer thickness in the nanometer range, and versatility. In addition to commonly used tuning parameters such as pH, ionic strength, and choice of polyelectrolytes, combining two different PEMs in a so-called asymmetric PEM enables further optimisation of the selective membrane layer. Experimental characterisation of these PEMs is complex, and therefore, the knowledge of the exact layer structure is limited. In this work, we combine filtration experiments and theoretical transport models to describe the effective structure of an asymmetric PEM made of a bottom layer of poly(allylamine hydrochloride) (PAH)/poly(sodium 4-styrenesulfonate) (PSS) and a top layer of PAH/poly(acrylic acid) (PAA). Obtained membrane properties suggest the formation of a distinct layer structure with individual layer properties close to the single symmetric PEM after a minimum number of layers. There is, however, a fundamental difference in the retention of salt and polyethylene glycol molecules. While salt retention properties of the asymmetric PEM are stable already after only one bilayer of PAH/PAA, a gradual transition in the retention of polyethylene glycol molecules from the more open PAH/PSS system to the dense PAH/PAA system is observed. This is attributed to the different exclusion mechanisms dominating solute transport (size- vs charge-based). A gradual decrease in molecular weight cut-off (MWCO) with increasing bilayer (BL) number of PAH/PAA is observed, resulting in a minimum MWCO of around 120g mol−1 after 5 BLs. Theoretical transport models assuming an ideal layer structure in series predict this value as well. At the same time, high mono- to divalent salt selectivity is observed after 1 BL of PAH/PAA already. Single salt retention of NaCl is around 20% versus Na2SO4, MgCl2 and MgSO4 being above 95%. Although this trend is qualitatively predicted by the theoretical transport model, again assuming ideal layers in series, deviations indicate variations in charge distribution within these layers.

On the Mechanical Buckling Analysis of FG-GRC Laminated Plates with Temperature-Dependent Material Properties Using Isogeometric Approach

In this paper, the isogeometric method is developed to study mechanical buckling behavior of nanocomposite plates reinforced by graphene sheets with temperature-dependent (TD) material properties in thermal environment. The plate is separately subjected to in-plane uniaxial, biaxial and shear loadings. It is assumed that the plate has different number of layers. By considering different volume fraction for each layer of graphene sheets, different functionally graded (FG) patterns of graphene sheets may be achieved. Furthermore, in some cases, it is considered that more than one FG patterns exist along the plate thickness. The energy statement of the plate is obtained using a logarithmic higher-order shear deformation theory (HSDT). Then, the isogeometric method is used to establish the desired eigenvalue problem. The comparison and convergence studies are presented for a wide range of numerical examples in all considered cases to show the correctness and ability of the solution. Afterwards, by presenting a set of numerical examples, the effects of plate significant parameters on the critical buckling load of the plate are examined. It is shown that the highest critical buckling loads occur when the plate has the minimum number of layers.

Three-Layer Compact K/Ka Feed Chain Using Enhanced Ortho-Mode Junction

This letter presents a compact dual-band dual circular polarization feed chain consisting of only three metallic layers. The minimum number of layers are accomplished by applying an asymmetric architecture based on an enhanced ortho-mode junction (OMJ) with single quad-ridge waveguide machined in one layer. This solution achieves the compactness of the feed chain while maintaining outstanding electrical performance. The feed chain in the band of 17.7-21.2 GHz and 28.2-31.4 GHz is designed and measured. The measured results show that the reflection coefficient of the common port is less than 19.2 dB, the reflection coefficient in both operative bands is better than 25.5 dB, the left-hand circular polarization (LHCP) to right-hand circularly polarization (RHCP) isolation is higher than 20.3 dB, the transceiver isolation is higher than 40 dB, and the axial ratio is below 0.35 dB. The measured results are in great agreement with predicted ones.

EVALUATION OF PERFORMANCE EFFICIENCY OF PACKING A GROUP OF PRODUCTS IN THE WORKPLACE OF ADDITIVE MACHINE USING A GENETIC ALGORITHM

Research results of possibilities of packing a group of 3D-models of products in a layered build space using a genetic algorithm are presented. It is proposed to determine the efficiency of the optimization problem of rational arrangement of 3D-models group in the workspace of additive machines depending on the number of loaded products. Condition for efficient use of the layered build workspace is the minimum number of layers per product and the largest relative filling. Such criteria are important, for example, for SLS/SLM technologies. Examples of evaluation based on the analysis of derived voxel 3D model of the workspace with located products are considered. Industrial products with different geometrical complexity were selected as test 3D models. This approach allowed to perform a comparative analysis of the results depending on the design features of products. The practical realization was performed in the subsystem of packing 3D-models in a workspace, which is part of the technological preparation system for the manufacture of complex products by additive methods. This system was developed at the Department of "Integrated Technologies of Mechanical Engineering" named after M. Semko of NTU "KhPI".

Cost-Effectively 3D-Printed Rigid and Versatile Interpenetrating Polymer Networks.

Versatile acrylate–epoxy hybrid formulations are becoming widespread in photo/thermal dual-processing scenarios, especially in 3D printing applications. Usually, parts are printed in a stereolithography or digital light processing (DLP) 3D printer, after which a thermal treatment would bestow the final material with superior mechanical properties. We report the successful formulation of such a hybrid system, consisting of a commercial 3D printing acrylate resin modified by an epoxy–anhydride mixture. In the final polymeric network, we observed segregation of an epoxy-rich phase as nano-domains, similar to what was observed in a previous work. However, in the current work, we show the effectiveness of a coupling agent added to the formulation to mitigate this segregation for when such phase separation is undesired. The hybrid materials showed significant improvement of Young’s modulus over the neat acrylate. Once the flexible, partially-cured material was printed with a minimal number of layers, it could be molded into a complex form and thermally cured. Temporary shapes were readily programmable on this final material, with easy shape recovery under mild temperatures. Inspired by repairable 3D printed materials described recently, we manufactured a large object by printing its two halves, and then joined them covalently at the thermal cure stage with an apparently seamless union.

High Sensitivity to Salinity-Temperature Using One-Dimensional Deformed Photonic Crystal

This paper aims to theoretically study the concept of a photonic salinity and temperature sensor according to a deformed one-dimensional photonic structure. The fundamental capability of the proposed sensor is studied. Simultaneously we search to optimize the thickness of the structure and to get the maximum salinity and temperature sensitivity. The structure is constructed by alternating layers of TiO2 and fused-silica P times. In the middle of the structure, a cavity containing seawater is inserted to measure its salinity and temperature. The transfer matrix method (TMM) is used to simulate the wave-transmittance spectra. It is shown that the quality factor (Q-factor) of the resonance peaks depends on the number (P) of layers. After that, the thickness of the layers is deformed by changing the deformation degree (h). The parameters P and h are optimized to get the maximal Q-factor with the minimal number of layers and structure thickness. The best sensitivity SS of the proposed salinity sensor is 558.82 nm/RFIU with a detection limit of 0.0034 RFIU. In addition, the best sensitivity ST of the designed temperature sensor is 600 nm/RFIU with a detection limit of 0.0005 RFIU.

A scalable algorithm for the optimization of neural network architectures

We propose a new scalable method to optimize the architecture of an artificial neural network. The proposed algorithm, called Greedy Search for Neural Network Architecture, aims to determine a neural network with minimal number of layers that is at least as performant as neural networks of the same structure identified by other hyperparameter search algorithms in terms of accuracy and computational cost. Numerical results performed on benchmark datasets show that, for these datasets, our method outperforms state-of-the-art hyperparameter optimization algorithms in terms of attainable predictive performance by the selected neural network architecture, and time-to-solution for the hyperparameter optimization to complete.

Application of the residue number system to reduce hardware costs of the convolutional neural network implementation

Convolutional neural networks are a promising tool for solving the problem of pattern recognition. Most well-known convolutional neural networks implementations require a significant amount of memory to store weights in the process of learning and working. We propose a convolutional neural network architecture in which the neural network is divided into hardware and software parts to increase performance and reduce the cost of implementation resources. We also propose to use the residue number system (RNS) in the hardware part to implement the convolutional layer of the neural network. Software simulations using Matlab 2018b showed that convolutional neural network with a minimum number of layers can be quickly and successfully trained. The hardware implementation of the convolution layer shows that the use of RNS allows to reduce the hardware costs on 7.86%–37.78% compared to the two’s complement implementation. The use of the proposed heterogeneous implementation reduces the average time of image recognition by 41.17%.

Predicting the structures and properties of few-layer two-dimensional (1 1 0)-oriented BN nanosheets: First-principles calculations

Based on first-principles calculations, we investigate the structural, electronic and optical properties of few-layer two-dimensional (2D) (1 1 0)-oriented BN nanosheets in relation to the layer number (n). After structural optimization, for n = 1 and n = 2, the buckling 2D BN nanosheets are reconstructed into planar sheets. With an increasing layer number n ≥ 3, the original bulk c-BN structures are energetically favorable to being maintained, and their properties tend to stabilize. It is demonstrated that the minimum number of layers that can satisfy the cubic phase structure of the (1 1 0) surface is 3; the thickness is approximately 5 Å. What is more interesting is that there are characteristics of the electronic and optical properties that are distinct from those of bulk c-BN material; the band gap and static refractive index values decrease to nearly half of those of the bulk material when the dimension changes from 3D to 2D. The development of novel structures in BN material fields is of great significance for both basic science and practical industrial applications.

APPLICATION OF KOHONEN SELF-ORGANIZING MAP TO SEARCH FOR REGION OF INTEREST IN THE DETECTION OF OBJECTS

Today, there is a serious need to improve the performance of algorithms for detecting objects in images. This process can be accelerated with the help of preliminary processing, having found areas of interest on the images where the probability of object detection is high. To this end, it is proposed to use the algorithm for distinguishing the boundaries of objects using the Sobel operator and Kohonen self-organizing maps, described in this paper and shown by the example of determining zones of interest when searching and recognizing objects in satellite images. The presented algorithm allows 15–100 times reduction in the amount of data arriving at the convolutional neural network, which provides the final recognition. Also, the algorithm can significantly reduce the number of training images, since the size of the parts of the input image supplied to the convolution network is tied to the image scale and equal to the size of the largest recognizable object, and the object is centered in the frame. This allows to accelerate network learning by more than 5 times and increase recognition accuracy by at least 10 %, as well as halve the required minimum number of layers and neurons of the convolutional network, thereby increasing its speed.

Chemical-Deposited Biaxially Textured Oxides YBCO, Y2O3 and LZO as Functional Layers for 2G HTSC Tapes

Abstract—2G HTSC tapes YBCO/LZO/Ni-5W and YBCO/Y2O3/LZO/Ni-5W with the minimum number of layers are obtained. Heteroepitaxial structures (layers La2Zr2O7, Y2O3 and YBa2Cu3O7) are formed by chemical deposition from a solution and chemical vapor deposition. Their textural quality is confirmed by XRD. The influence of the layers on each other is shown. Many factors for improving the quality of the 2G HTSC tapes obtained by chemical methods are considered.

Hardware implementation of a convolutional neural network using calculations in the residue number system

Modern convolutional neural networks architectures are very resource intensive which limits the possibilities for their wide practical application. We propose a convolutional neural network architecture in which the neural network is divided into hardware and software parts to increase performance and reduce the cost of implementation resources. We also propose to use the residue number system in the hardware part to implement the convolutional layer of the neural network for resource costs reducing. A numerical method for quantizing the filters coefficients of a convolutional network layer is proposed to minimize the influence of quantization noise on the calculation result in the residue number system and determine the bit-width of the filters coefficients. This method is based on scaling the coefficients by a fixed number of bits and rounding up and down. The operations used make it possible to reduce resources in hardware implementation due to the simplifying of their execution. All calculations in the convolutional layer are performed on numbers in a fixed-point format. Software simulations using Matlab 2017b showed that convolutional neural network with a minimum number of layers can be quickly and successfully trained. Hardware implementation using the field-programmable gate array Kintex7 xc7k70tfbg484-2 showed that the use of residue number system in the convolutional layer of the neural network reduces the hardware costs by 32.6% compared with the traditional approach based on the two’s complement representation. The research results can be applied to create effective video surveillance systems, for recognizing handwriting, individuals, objects and terrain.

INFLUENCE OF GEOMETRICAL PARAMETERS OF THE INDUCTOR AND ARMATURE ON THE INDICATORS OF A LINEAR PULSE ELECTROMECHANICAL CONVERTER OF AN ELECTRODYNAMIC TYPE

Purpose. The aim of the paper is to study the influence of geometrical parameters, namely, the number of layers and the cross section of the copper tire of the inductor and the armature coils on the power and speed indicators of a linear pulse electromechanical converter (LPEC) of an electrodynamic type. Methodology. On the basis of the developed chain mathematical model, recurrent relations are obtained for the calculation of interconnected electromagnetic, mechanical and thermal processes of LPEC of an electrodynamic type. The effect of the thickness of a square copper tire and the number of its layers in the inductor and armature coils on the characteristics and characteristics of electrodynamic LPEC is investigated. It is these parameters that determine the number of turns and the axial height of the coils with limited radial dimensions. Results. The influence of the geometrical parameters of the inductor and the armature coils with limited radial dimensions on the electrical and mechanical characteristics of LPEC of an electrodynamic type is established. It has been established that with an increase in the thickness of a rectangular cross-section of copper tire from 1 to 2.5 mm, an increase in the amplitude and pulse of electrodynamic forces (EF) occurs. However, the maximum speed of the armature is the highest at LPEC wound with a 1.5 mm thick tire. The highest efficiency value is demonstrated by LPEC, in which the inductor and armature coils are wound with a 2 mm thick tire. With an increase in the number of layers of the inductor coil tire, the amplitude of the EF decreases significantly, and the magnitude of the EF pulse decreases slightly. As a result, the maximum armature speed, efficiency and temperature rise of the coils are reduced. Originality. It is established that the largest amplitude of the EF is realized in LPEC with the minimum number of layers of tires of the inductor and armature coils. The largest value of the pulse EF occurs when the maximum number of layers of the inductor and the armature. In this case, the largest values of the amplitude and pulse of the EF occur under the condition that the number of tire layers of the inductor and the armature coils are the same. Practical value. It has been established that the greatest efficiency 21.82 % is realized in LPEC, in which the number of tire layers is 2 mm thick with inductor and armature coils are 4. A catapult model for launching an unmanned aerial vehicle was made and tested on the basis of LPEC of an electrodynamic type.

High efficient ultra-broadband anti-reflection coating on silicon for infrared applications

In this work, we describe an efficient design and fabrication process of anti-reflection (AR) coating over 1400 nm spectral interval from 3.4 μm to 4.8 μm at mid wave infrared (MWIR) on silicon optics with only 3 layers which have overall physical thickness less than 1 μm. Multi-layer AR coating consisting of aluminum oxide (Al2O3) and germanium (Ge) layers as low (nL) and high (nH) index materials respectively are coated by plasma assisted e-beam evaporation on silicon substrates with well optimized process parameters. AR coating on silicon is the first one reported by exhibiting record average transmission more than 99.7% over 1400 nm broadband spectral interval in mid-wave infrared (MWIR) with minimum number of layers. Moreover, number of layers and overall thickness for our design enables a cost effective solution for low process cycle in mass production of AR coated silicon optics.

Deep Transfer Learning Models for Medical Diabetic Retinopathy Detection.

Introduction:Diabetic retinopathy (DR) is the most common diabetic eye disease worldwide and a leading cause of blindness. The number of diabetic patients will increase to 552 million by 2034, as per the International Diabetes Federation (IDF).Aim:With advances in computer science techniques, such as artificial intelligence (AI) and deep learning (DL), opportunities for the detection of DR at the early stages have increased. This increase means that the chances of recovery will increase and the possibility of vision loss in patients will be reduced in the future.Methods:In this paper, deep transfer learning models for medical DR detection were investigated. The DL models were trained and tested over the Asia Pacific Tele-Ophthalmology Society (APTOS) 2019 dataset. According to literature surveys, this research is considered one the first studies to use of the APTOS 2019 dataset, as it was freshly published in the second quarter of 2019. The selected deep transfer models in this research were AlexNet, Res-Net18, SqueezeNet, GoogleNet, VGG16, and VGG19. These models were selected, as they consist of a small number of layers when compared to larger models, such as DenseNet and InceptionResNet. Data augmentation techniques were used to render the models more robust and to overcome the overfitting problem.Results:The testing accuracy and performance metrics, such as the precision, recall, and F1 score, were calculated to prove the robustness of the selected models. The AlexNet model achieved the highest testing accuracy at 97.9%. In addition, the achieved performance metrics strengthened our achieved results. Moreover, AlexNet has a minimum number of layers, which decreases the training time and the computational complexity.

Direct exfoliation of graphite and its Raman spectroscopic study

Direct exfoliating is a safer method for graphene preparation. However, it is important to synthesize a graphene with minimum structural defects since they will influence the properties of the devices. This paper focuses on two types of graphite exfoliation techniques, electrochemical and sonication, for the preparation of graphene oxides. The quality of the graphene oxides was studied using Raman Spectroscopy while the morphology was investigated by field emission scanning electron microscopy (FESEM). The results suggest that the graphene oxides synthesized using the electrochemical technique has the minimum number of layers while graphene oxides obtained from the sonication technique has the minimum defects.

One-shot domain adaptation in multiple sclerosis lesion segmentation using convolutional neural networks.

In recent years, several convolutional neural network (CNN) methods have been proposed for the automated white matter lesion segmentation of multiple sclerosis (MS) patient images, due to their superior performance compared with those of other state-of-the-art methods. However, the accuracies of CNN methods tend to decrease significantly when evaluated on different image domains compared with those used for training, which demonstrates the lack of adaptability of CNNs to unseen imaging data. In this study, we analyzed the effect of intensity domain adaptation on our recently proposed CNN-based MS lesion segmentation method. Given a source model trained on two public MS datasets, we investigated the transferability of the CNN model when applied to other MRI scanners and protocols, evaluating the minimum number of annotated images needed from the new domain and the minimum number of layers needed to re-train to obtain comparable accuracy. Our analysis comprised MS patient data from both a clinical center and the public ISBI2015 challenge database, which permitted us to compare the domain adaptation capability of our model to that of other state-of-the-art methods. In both datasets, our results showed the effectiveness of the proposed model in adapting previously acquired knowledge to new image domains, even when a reduced number of training samples was available in the target dataset. For the ISBI2015 challenge, our one-shot domain adaptation model trained using only a single case showed a performance similar to that of other CNN methods that were fully trained using the entire available training set, yielding a comparable human expert rater performance. We believe that our experiments will encourage the MS community to incorporate its use in different clinical settings with reduced amounts of annotated data. This approach could be meaningful not only in terms of the accuracy in delineating MS lesions but also in the related reductions in time and economic costs derived from manual lesion labeling.