Neural Network Tool Research Articles

ScGNN 2.0: a graph neural network tool for imputation and clustering of single-cell RNA-Seq data.

Gene expression imputation has been an essential step of the single-cell RNA-Seq data analysis workflow. Among several deep-learning methods, the debut of scGNN gained substantial recognition in 2021 for its superior performance and the ability to produce a cell-cell graph. However, the implementation of scGNN was relatively time-consuming and its performance could still be optimized. The implementation of scGNN 2.0 is significantly faster than scGNN thanks to a simplified close-loop architecture. For all eight datasets, cell clustering performance was increased by 85.02% on average in terms of adjusted rand index, and the imputation Median L1 Error was reduced by 67.94% on average. With the built-in visualizations, users can quickly assess the imputation and cell clustering results, compare against benchmarks and interpret the cell-cell interaction. The expanded input and output formats also pave the way for custom workflows that integrate scGNN 2.0 with other scRNA-Seq toolkits on both Python and R platforms. scGNN 2.0 is implemented in Python (as of version 3.8) with the source code available at https://github.com/OSU-BMBL/scGNN2.0. Supplementary data are available at Bioinformatics online.

Numerical prediction of microstructure and hardness for low carbon steel wire Arc additive manufacturing components

The objective of this research is to define an optimized strategy for wire arc additive manufacturing (WAAM) process with a numerical tool. In this paper, a three-dimensional scaled steel plate model (120 mm × 24 mm × 6 mm) is built for numerical simulations to replace the expensive physical WAAM experiments. A series of thermo-metallo-mechanical analyses is conducted by using ABAQUS CAE user subroutines in which, thermal, metallurgical and mechanical models for gas metal arc welding (GMAW) process are implemented. Groups of variables in WAAM process namely wire melting current, voltage and torch moving speeds, as well as interval layer cooling coefficients, are chosen as input parameters of the artificial neural network (ANN). The two targeted values are hardness and ultimate strength of the steel plate model, which are obtained by fully coupled thermo-metallo-mechanical simulations or by quick results from commercial software JmatPro, and by real measurements on the test samples of WAAM experimental plates. In this primary study, Taguchi algorithm is used for data preparations of the numerical experiments. The tests are fulfilled with Neural Networks (NN) tool in MATLAB. A typical three-layer feedforward network is used for general works. The developed models show good prediction capability for hardness and ultimate strength. The ANN algorithm presents high potential in bidirectional modeling to develop the WAAM strategy tool. It is possible to realize the inverse modeling from properties obtained by fractions of phase volume to WAAM processing parameters.

Artificial Neural Network and Weighted Arithmetic Indexing Approach for Surface Water Quality Assessment of the Brahmani River

<p>An approach was used in this study to relate the predicted and calculated water quality index (WQI) of the Brahmani River. The WQI was predicted using an artificial neural network (ANN) tool, and the weighted Arithmetic Index technique was used to calculate the WQI (WAI). The WQI is calculated using physicochemical parameters as input data. Pollution Control Board (India) data was utilised to train and evaluate the model, as well as to forecast WQI. The ANN model is trained using the feed-forward back-propogation approach. 70 percent of the data was used for training, whereas 30 percent was used for testing and validation (15 percent) (15 percent). The regression coefficients for all of the stations were greater than 0.9, indicating that ANN modelling produced successful results. For all stations, the average percent of variance between anticipated and computed WQI is 8.63 percent. According to the findings of this study, the ANN model may be useful for predicting the WQI of both surface and groundwater.</p>

Biomedical Diagnosis of Leukemia Using a Deep Learner Classifier.

Leukemia cancer is the most common type of cancer that occurs in childhood. The most common types are acute lymphocytic leukemia (ALL) and acute myelogenous leukemia (AML) which affect children and adults, respectively. Several health issues occur due to these cancers. Leukemia affects the bone marrow or the lymph nodes. Leukemia produces abnormal white blood cells via the bone marrow system. The affected white blood cells are unable to perform their tasks properly. Detecting leukemia usually requires taking a blood smear from a patient and working with expert hematologists who analyze the smear with a microscope. In this paper, a method to detect ALL and AML using a deep learner classifier is developed and proposed. The method detects both types, determines their severity, and creates a message that recommends next steps to patients. This approach works based on image segmentation and a convolutional neural network (CNN) tool called AlexNet. The obtained results from the proposed approach and using MATLAB reached more than 98% accuracy. The margin exists because several operations are needed to fully detect the blood cancer. A dataset of leukemia from the Kaggle site is used to test the developed method and illustrate its effectiveness. This dataset is C-NMC_Leukemia, and it consists of nearly 10 GB worth of 15,000 images. A confusion matrix of testing images is provided to prove the correctness of the presented approach. Furthermore, a comparative analysis between the proposed algorithm and some works from the literature is presented. This analysis compares the method used to extract features, the classifier that is utilized, the accuracy, the precision, and the recall. The obtained results indicate that the proposed method outperforms other works and produces better results.

An artificial neural network tool to support the decision making of designers for environmentally conscious product development

The consideration of sustainability aspects in initial stages of product development is understood as an effective approach to plan a sustainable product life cycle. It can be achieved by integrating the sustainability considerations into decision making of companies in early design phases. This study aims to develop an Artificial Intelligence (AI) tool that can assist the designers in their decision making to choose environmentally benign design parameters of products. The proposed tool is based on an Artificial Neural Network (ANN) model which takes the life cycle design parameters (viz. size of product, density of material, manufacturing process, transport mode and recyclability) as inputs and provides the corresponding outputs in terms of ‘carbon footprint’ and ‘life cycle cost’ of a product. These outputs assist the designers to realize the trade-off between the environmental load and cost effectiveness of a design alternative. Thus, it enables the decision making of companies to select a more sustainable design. A Graphical User Interface (GUI) is developed for the AI tool so that the designers can efficiently use this tool. The results predicted by the proposed tool are compared with the results obtained through the life cycle assessment carried out by using GaBi 9.2. The comparison shows that the results predicted by the tool have a reasonable accuracy of more than 90% which is significant, especially in the design stages of environmentally conscious product development. Also, the time efficiency of the tool to compute the environmental impact was compared with that of GaBi 9.2 by using a T-test. Results showed that the time efficiency of the proposed AI tool is significantly higher than that of GaBi 9.2.

Study on the Application of Improved Audio Recognition Technology Based on Deep Learning in Vocal Music Teaching

As one of the hotspots in music information extraction research, music recognition has received extensive attention from scholars in recent years. Most of the current research methods are based on traditional signal processing methods, and there is still a lot of room for improvement in recognition accuracy and recognition efficiency. There are few research studies on music recognition based on deep neural networks. This paper expounds on the basic principles of deep learning and the basic structure and training methods of neural networks. For two kinds of commonly used deep networks, convolutional neural network and recurrent neural network, their typical structures, training methods, advantages, and disadvantages are analyzed. At the same time, a variety of platforms and tools for training deep neural networks are introduced, and their advantages and disadvantages are compared. TensorFlow and Keras frameworks are selected from them, and the practice related to neural network research is carried out. Training lays the foundation. Results show that through the development and experimental demonstration of the prototype system, as well as the comparison with other researchers in the field of humming recognition, it is proved that the deep-learning method can be applied to the humming recognition problem, which can effectively improve the accuracy of humming recognition and improve the recognition time. A convolutional recurrent neural network is designed and implemented, combining the local feature extraction of the convolutional layer and the ability of the recurrent layer to summarize the sequence features, to learn the features of the humming signal, so as to obtain audio features with a higher degree of abstraction and complexity and improve the performance of the humming signal. The ability of neural networks to learn the features of audio signals lays the foundation for an efficient and accurate humming recognition process.

Artificial neural network combined with damage parameters to predict fretting fatigue crack initiation lifetime

The fretting problem occurs at two contact surfaces sustaining small relative displacement, and it reduces the fatigue lifetime dramatically. Estimating accurate fretting fatigue lifetime plays an important role in engineering applications. Due to the complicated stress state, and high-stress gradient in the contact surface, the average methods are necessary to obtain the precise lifetime, but the critical distance for the average zone is difficult to estimate. In this work, Artificial Neural Network (ANN) tool combined with damage parameters is proposed to determine the optimal critical distance for different fretting conditions. This tool can also be used to accurately predict the crack initiation lifetime. The fretting fatigue lifetimes calculated by using this approach have shown good agreement with experimental results from literatures. In addition, rough estimates of critical distance for different cases are made based on the numerical results.

Neural network based uncertainty and sensitivity evaluation of electrical resistivity tomography for improved subsurface imaging

Assessment of subsurface status by resistivity technique, being an indirect approach, is pretended to be a strategic factor. Projection of full proof confirmation in this domain is always a challenge and hence outcomes are expressed in possibilities. Intervene of mathematical interpretation on resistivity data generated in the field by different arrays would offer a better choice in building-up the possibility of projecting the actual status. Thus, a study of Wenner-Schlumberger (WS), dipole–dipole (DD) and combined inversion (CI) data of three parallel profiles have been conducted, as a whole, for old and abandoned shallow depth coal mine workings in Jharia coalfield. The study recapitulates influence of sensitivity and uncertainty with depth, apart from resistivity. Statistical significance of the data has been evaluated inclusive of their inter-relationship. PCA presented an encouraging relation of sensitivity with depth. The comprehensible approach of mathematical interpretation helps in cracking a problem of uncertain prediction. Sensitivity and the extent of uncertainty are the parameters to build a strong foundation for evaluating the degree of confidence in prediction accuracy. Artificial Neural Network (ANN) tool has been used to understand the relative importance of sensitivity and uncertainty with depth. The weightage of sensitivity has been observed to be on upper side with respect to uncertainty. The importance of configuration of resistivity survey array has been emphasized based on sensitivity.

Multi-Angle Projection Based Blind Omnidirectional Image Quality Assessment

Most of the existing blind omnidirectional image quality assessment (BOIQA) methods are based on data-driven approach where the end-to-end neural network or deep learning tools are mainly used for feature extraction. However, it usually lacks interpretability and is difficult to discover the perceptual mechanism behind. In this paper, from the perspective of perception modeling, we propose a novel multi-angle projection based BOIQA (MP-BOIQA) method. Considering the omnibearing and near eye display characteristics with head mounted display, multiple color cubemap projection images with respect to different viewpoints are grouped as the color omnidirectional distortion (COD) units so as to simulate the user’s viewing behavior in subjective quality assessment. In the designed multi-angle projection based feature extractor, tensor decomposition is implemented on each COD unit for dimensionality reduction, and piecewise exponential fitting is used to get the distribution of mean subtracted contrast normalized coefficients of the unit’s feature matrices in tensor domain. Finally, the extracted features are pooled with random forest. The experimental results on three omnidirectional image quality datasets show that the MP-BOIQA method can deliver highly competitive performance compared with some representative full-reference quality assessment methods, as well as some state-of-the-art BOIQA methods.

Detection of photo forgery performed by tools of neural networks

Detection of photo forgery performed by tools of neural networks

GoTube: Scalable Statistical Verification of Continuous-Depth Models

We introduce a new statistical verification algorithm that formally quantifies the behavioral robustness of any time-continuous process formulated as a continuous-depth model. Our algorithm solves a set of global optimization (Go) problems over a given time horizon to construct a tight enclosure (Tube) of the set of all process executions starting from a ball of initial states. We call our algorithm GoTube. Through its construction, GoTube ensures that the bounding tube is conservative up to a desired probability and up to a desired tightness. GoTube is implemented in JAX and optimized to scale to complex continuous-depth neural network models. Compared to advanced reachability analysis tools for time-continuous neural networks, GoTube does not accumulate overapproximation errors between time steps and avoids the infamous wrapping effect inherent in symbolic techniques. We show that GoTube substantially outperforms state-of-the-art verification tools in terms of the size of the initial ball, speed, time-horizon, task completion, and scalability on a large set of experiments. GoTube is stable and sets the state-of-the-art in terms of its ability to scale to time horizons well beyond what has been previously possible.

Damage assessment of aircraft wing subjected to blast wave with finite element method and artificial neural network tool

Damage assessment of the wing under blast wave is essential to the vulnerability reduction design of aircraft. This paper introduces a critical relative distance prediction method of aircraft wing damage based on the back-propagation artificial neural network (BP-ANN), which is trained by finite element simulation results. Moreover, the finite element method (FEM) for wing blast damage simulation has been validated by ground explosion tests and further used for damage mode determination and damage characteristics analysis. The analysis results indicate that the wing is more likely to be damaged when the root is struck from vertical directions than others for a small charge. With the increase of TNT equivalent charge, the main damage mode of the wing gradually changes from the local skin tearing to overall structural deformation and the overpressure threshold of wing damage decreases rapidly. Compared to the FEM-based damage assessment, the BP-ANN-based method can predict the wing damage under a random blast wave with an average relative error of 4.78%. The proposed method and conclusions can be used as a reference for damage assessment under blast wave and low-vulnerability design of aircraft structures.

Gamma-ray energy spectrum unfolding of plastic scintillators using artificial neural network

In this study, the unfolding of the plastic scintillator spectrum was undertaken using the artificial neural networks tools of MATLAB. To this purpose, the response matrix of the plastic scintillator was generated for 145 energy groups and in 512 pulse-height channels using the MCNPX2.6 code. The results confirmed that the relative error in the gamma-ray energy unfolding with artificial neural networks is less than 3.8%.

FC048: New Tool to Predict the Clinical Course and Renal Failure in Patients with Immunoglobulin a Nephropathy

Abstract BACKGROUND AND AIMS Idiopathic Immunoglobulin A nephropathy (IgAN) is the most common biopsy-proven glomerulonephritis in the world. Approximately 40% of IgAN patients reach renal failure (RF) 20 years after their kidney biopsy. The high prevalence of RF shows that IgAN has a significant economic impact in the countries because renal replacement therapy is costly. Moreover, the disease's onset in the second and third decades of life represents a social challenge because patients are typically very active and highly productive in the workplace. This challenge is one more reason to move on the prediction of the clinical course and RF in IgAN patients at the time of the kidney biopsy and during the follow-up. We developed an artificial neural network (ANN) tool (DialCheck 1.0) based on seven variables and the histological score of the kidney biopsy to predict RF in IgAN patients at the time of kidney biopsy [1]. METHOD We have recently developed a new tool which consists of a set of ANN-powered models that combine temporally accurate observations for fine-grained features and leverage state-of-the-art deep neural network techniques to forecast the patient's clinical evolution. RESULTS A cohort of 948 IgAN patients, of whom the clinical course of the disease was known, was used to develop the new tool that predicts the dynamics of age and disease laboratory parameters (serum creatinine, daily proteinuria), blood pressure, histological score of the kidney biopsy and the RF. The system was designed to help the physicians give a broader spectrum of information regarding the patient and the potential clinical development of IgAN and outcomes. In detail, the model computes a latent dynamic representation of the patient and predicts a prospective clinical picture of the patient and the probability of RF. The tool with an accuracy of >80% will be tested in an independent retrospective cohort of 454 IgAN patients and in a prospective multicenter randomized clinical study in which 426 IgAN patients will be enrolled in two different groups based on the type of kidney lesions (active and chronic renal lesions). CONCLUSION We have a new tool (DialCheck 2.0), based on ANN, that may predict the outcome and the RF in IgAN patients. Moreover, it may help the physician to analyze the long-term response to therapy.

Data-driven analysis on axial strength of GFRP-NSC columns based on practical artificial neural network tool

Previous research is deficient in the theoretical estimation of the axial load-carrying strength (ALCS) of glass fiber reinforced polymer (GFRP) composite columns. This study aims to calculate the ALCS of GFRP normal strength concrete (NSC) columns (GFRP-NSC) using two different approaches; empirical modeling and artificial neural networks (ANNs). The geometric, material, and testing details of 250 GFRP-NSC columns were collected from the previous studies. An initial assessment of the previous strength models for the ALCS of GFRP-NSC columns was accomplished over the constructed database. General regression and curve-fitting were used for suggesting an improved empirical model. The various hidden layers and neurons were calibrated to find an improved ANN model. Both new empirical and ANN models showed higher accuracy than the previous model for capturing the ALCS with R2 = 0.740, and R2 = 0.882, individually. A comparison of the predictions through one-way analysis of variance (ANOVA) at a 5% significance level depicts that the proposed models presented a higher accuracy than the previous models for capturing the ALCS of GFRP-NSC columns. Additionally, a detailed parametric investigation of GFRP-NSC members was carried out using the suggested empirical model to explore the influence of different variables of studied members.

WalkIm: Compact image-based encoding for high-performance classification of biological sequences using simple tuning-free CNNs.

The classification of biological sequences is an open issue for a variety of data sets, such as viral and metagenomics sequences. Therefore, many studies utilize neural network tools, as the well-known methods in this field, and focus on designing customized network structures. However, a few works focus on more effective factors, such as input encoding method or implementation technology, to address accuracy and efficiency issues in this area. Therefore, in this work, we propose an image-based encoding method, called as WalkIm, whose adoption, even in a simple neural network, provides competitive accuracy and superior efficiency, compared to the existing classification methods (e.g. VGDC, CASTOR, and DLM-CNN) for a variety of biological sequences. Using WalkIm for classifying various data sets (i.e. viruses whole-genome data, metagenomics read data, and metabarcoding data), it achieves the same performance as the existing methods, with no enforcement of parameter initialization or network architecture adjustment for each data set. It is worth noting that even in the case of classifying high-mutant data sets, such as Coronaviruses, it achieves almost 100% accuracy for classifying its various types. In addition, WalkIm achieves high-speed convergence during network training, as well as reduction of network complexity. Therefore WalkIm method enables us to execute the classifying neural networks on a normal desktop system in a short time interval. Moreover, we addressed the compatibility of WalkIm encoding method with free-space optical processing technology. Taking advantages of optical implementation of convolutional layers, we illustrated that the training time can be reduced by up to 500 time. In addition to all aforementioned advantages, this encoding method preserves the structure of generated images in various modes of sequence transformation, such as reverse complement, complement, and reverse modes.

Fractional Order Neural Transient Modeling of Primary Circuit of ACP1000 Based Nuclear Power Plant in LabVIEW

The primary circuit of the nuclear power plant is the most advanced and sophisticated loop of the Advanced Chinese Pressurized Water Reactor (ACP1000). The primary circuit is composed of most technologically advanced nuclear systems and controllers. In this research work, closed loop dynamics of primary circuit (CLPC) of ACP1000 based nuclear power plant is identified. The closed loop dynamics is comprised of highly nonlinear coupled sevencontrol systems. The turbine power, pressurizer temperature, cold leg temperature, hot leg temperature, coolant average temperature and feed water flow are the selected parameters of interest as inputs while neutron power, reactor coolant pressure, pressurizer level, steam generator pressure, steam generator level and steam generator flow as outputs. Therefore, a closed loop multi-input multi-out (MIMO) is configured. The control oriented closed loop dynamics of the primary circuit of ACP1000 is estimated by state-of-the-art novel fractional order neural network (FO-ANN) tool developed in LabVIEW. The parameters of FO-ANN of CLPC (FO-ANN-CLPC) are optimized using fractional order backpropagation (FO-BP) algorithm. The performance of FO-ANN-CLPC is tested and validated in transient conditions and the proposed model predicted the desired reactor power with minimizing error function. The robust performance of the proposed closed loop model is evaluated by dynamic simulation for a prescribed turbine load power increase transient from 20 % to 100 % and validated against reactor power and behaviour of various thermal hydraulics parameters are observed and analyzed.

Dense, deep learning-based intracranial aneurysm detection on TOF MRI using two-stage regularized U-Net

Background and purposeThe prevalence of unruptured intracranial aneurysms in the general population is high and aneurysms are usually asymptomatic. Their diagnosis is often fortuitous on MRI and might be difficult and time consuming for the radiologist. The purpose of this study was to develop a deep learning neural network tool for automated segmentation of intracranial arteries and automated detection of intracranial aneurysms from 3D time-of-flight magnetic resonance angiography (TOF-MRA). Materials and methods3D TOF-MRA with aneurysms were retrospectively extracted. All were confirmed with angiography. The data were divided into two sets: a training set of 24 examinations and a test set of 25 examinations. Manual annotations of intracranial blood vessels and aneurysms were performed by neuroradiologists. A double convolutional neuronal network based on the U-Net architecture with regularization was used to increase performance despite a small amount of training data. The performance was evaluated for the test set. Subgroup analyses according to size and location of aneurysms were performed. ResultsThe average processing time was 15 min. Overall, the sensitivity and the positive predictive value of the proposed algorithm were 78% (21 of 27; 95% CI: 62–94) and 62% (21 of 34; 95%CI: 46–78) respectively, with 0.5 FP/case. Despite gradual improvement in sensitivity regarding aneurysm size, there was no significant difference of sensitivity detection between subgroups of size and location. ConclusionsThis developed tool based on a double CNN with regularization trained with small dataset, enables accurate intracranial arteries segmentation as well as effective aneurysm detection on 3D TOF MRA.

Revealing prediction of perched cum off-centered wick solar still performance using network based on optimizer algorithm

A sincere effort has been made to engineer a perched cum off-centered wick solar still (PCWSS) in the present work. The daily average efficiency with hourly prediction distillate yield of the PCWSS has used those artificial neural networks (ANNs) tool with Harris Hawk’s Optimizes (HHOs) technique. HHO performance with ANN simulated as an optimal parameter to grab preyed. An experimental performance predicting the system's productivity is associated by dual supplementary mockups as vectors gadget, tradition ANN. HHO-ANN approach results are compared with the experimental observations (one year) of the solar still. Radial Basis Function (RBF) and Feed Forward (FF) have been used ANN structures to estimate hourly distillate yield and efficiency of the system is 59.78%. Evaluating the R2, RMSE, MRE, MAE, EC, OI, CRM analysis of prediction models was based on numerical error conditions. Optimized the analysis of PCWSS with a model as HHO-ANN used optimal parameter values has prediction accuracy associated with ANN and the competence for HHO. Annual analysis based on the HHO - ANN structures predicted the hourly distillate yield with mean error varying from 8.13% and 6.1%. The error for the monthly average prediction of distillate yield is from 0.95% to 1.12%, respectively. HHO – ANN has been used with the best accuracy in predicting the PCWSS invention associated with tangible experimental outcomes.

Discrimination between inrush and fault currents of transformers using Artificial Neural Network Tools

Discrimination between inrush and fault currents of transformers using Artificial Neural Network Tools