Implementation Of Artificial Neural Networks Research Articles

Efficient implementation of artificial neural networks on FPGAs using high-level synthesis and parallelism

Efficient implementation of artificial neural networks on FPGAs using high-level synthesis and parallelism

Implementation of Artificial Neural Networks as a Method for Early Detection of Tax Evasion Behavior in Indonesia

Research Aims: This study aims to analyze the ability of the Artificial Neural Network (ANN) model in early detection of tax evasion behavior in Indonesia. Design/methodology/approach: This study uses a qualitative method with a descriptive approach, the types of data used in this study are primary data and secondary data. Primary data is obtained from the results of interviews to obtain the information needed by the researcher. The interview informants came from the DGT sub-Directorate of Potential, Compliance and Tax Revenue with the initials K&H, tax consultants and taxpayers. while secondary data is collected through the DGT online website and through articles and journals on Artificial Intelligence (AI), Artificial Neural Networks (ANN), tax evasion, and financial information data on income, expenses, amount of assets, taxes payable) and behavioral indicators (tax compliance history, frequency of late reporting, significant changes in financial statements). The data analysis technique used is descriptive analysis, namely data collection, data reduction, data presentation and conclusion drawing Research Findings: The results of this study show that the use of a neural network model can be an effective tool to detect symptoms of tax evasion behavior early. This method allows tax authorities to monitor and identify suspicious activity more efficiently and effectively. Theoretical Contribution/Originality: This study makes a significant theoretical contribution by integrating the use of Artificial Neural Networks (ANN) in detecting tax evasion in Indonesia. The originality of this research lies in the application of ANN in the context of Indonesia's tax system, where the use of advanced technology such as ANN is still relatively new. The study also combines the pentagon's fraud theory, which adds elements of arrogance and competence in the analysis of tax evasion behavior, with a machine learning approach, creating a more holistic and innovative framework

Enhancing IoT Security through an Artificial Neural Network Approach

This study aims to fortify Internet of Things (IoT) security through the strategic implementation of Artificial Neural Networks (ANNs). With the rapid expansion of IoT devices, traditional security measures have struggled to cope with the dynamic and complex nature of these environments. ANNs, known for their adaptability, are explored as a promising solution to enhance security. The central objective is to significantly improve the accuracy of IoT security measures by optimizing ANN architectures. Using a curated dataset with key environmental parameters, the study evaluates three ANN models—Backpropagation Neural Network (BPNN), Multilayer Perceptron (MLP), and Long Short-Term Memory (LSTM). The evaluation metrics include accuracy, precision, recall, and F1-score across different train-test splits. Results show that LSTM consistently outperforms BPNN and MLP, demonstrating superior accuracy and the ability to capture temporal dependencies within IoT security data. Implications stress the importance of aligning model selection with specific application goals, considering factors like computational efficiency. In conclusion, this research contributes valuable insights into the practical implementation of ANNs for IoT security, guiding future optimization efforts and addressing real-world deployment challenges to safeguard sensitive data and ensure system resilience in the evolving IoT landscape.

Прогнозирование агропродовольственных экономических систем с использованием искусственных нейронных сетей

Abstract. The purpose of the study was to substantiate the applicability of the use of artificial neural networks to the forecasting of agro-food economic systems. Methods. The research is based on the use of elements of the interpretative method in a combination of genetic, structural, functional, complex, systemic, and empirical approaches. The scientific novelty it consists in systematization of algorithms for the implementation of artificial neural networks and substantiation of their applicability for forecasting agro-food economic systems, development of an algorithm and architecture for building a neural network based on multiple data on agricultural markets, substantiation of directions for improving information infrastructure at the firm level. Results. The authors systematized intuitive and formalized forecasting methods, justified the place of methods based on machine learning in this system. The advantages and disadvantages of using artificial neural networks for forecasting agri-food economic systems are considered in detail, the expediency of their use from the point of view of compliance with the principles of forecasting is justified. The analysis of the main types of artificial neural networks allowed us to conclude that the most promising for the implementation of forecasting tasks are competitive neural networks with a back propagation algorithm (LSTM and GRU). The main objectives of building models based on neural networks for use in forecasting economic systems are formulated, the basic provisions of the sequence and methods of deploying neural networks in the forecasting process in the agri-food market are developed, the key elements of the organization of the forecasting process in individual economic entities are considered, practical aspects of the possibility of using a mathematical algorithm for modeling agri-food systems are considered, as well as the conditions for improving the information infrastructure at the firm level in order to ensure the availability of data sources and technologies for their processing.

Neuro-computing analysis of model-based Casson hybrid nanofluid flow via three-dimensional radiative Riga plate with irregular heat source/sink

In this study, an analysis is conducted on a hybrid nanofluid using the Yamada-Ota and Hamilton-Crosser model in combination with the Casson fluid model. The hybrid nanofluid is analyzed in the context of a three-dimensional radiative Riga plate, which incorporates non-uniform heat source/sink effects. The thermal features and their behavior of the hybridized nanofluid is explored in the current investigation considering the flow over a radiative Riga plate with the impact of space and temperature dependent heat source/sink. The flow and thermal characteristics of the hybridized nanomaterial are presented for the consideration of Yamada-Ota & Hamilton Crosser models. Additionally, the analysis is enhanced by the implementation of Artificial neural network. Further the machine learning approach, a neuro-computing intelligent network is adopted for the computational techniques. The neural networks, which effectively uses the data and make predictions based on their acquired knowledge. The outcomes of the proposed analysis provide the flow behavior of the hybrid nanofluid. The effect of non-uniform heat source/sink of the Casson fluid model contributes to a greater hike in the fluid temperature. Moreover, the neuro-computing intelligent network shows efficient predictions of the involved thermal parameters.

Backpropagation-Based Learning Techniques for Deep Spiking Neural Networks: A Survey.

With the adoption of smart systems, artificial neural networks (ANNs) have become ubiquitous. Conventional ANN implementations have high energy consumption, limiting their use in embedded and mobile applications. Spiking neural networks (SNNs) mimic the dynamics of biological neural networks by distributing information over time through binary spikes. Neuromorphic hardware has emerged to leverage the characteristics of SNNs, such as asynchronous processing and high activation sparsity. Therefore, SNNs have recently gained interest in the machine learning community as a brain-inspired alternative to ANNs for low-power applications. However, the discrete representation of the information makes the training of SNNs by backpropagation-based techniques challenging. In this survey, we review training strategies for deep SNNs targeting deep learning applications such as image processing. We start with methods based on the conversion from an ANN to an SNN and compare these with backpropagation-based techniques. We propose a new taxonomy of spiking backpropagation algorithms into three categories, namely, spatial, spatiotemporal, and single-spike approaches. In addition, we analyze different strategies to improve accuracy, latency, and sparsity, such as regularization methods, training hybridization, and tuning of the parameters specific to the SNN neuron model. We highlight the impact of input encoding, network architecture, and training strategy on the accuracy-latency tradeoff. Finally, in light of the remaining challenges for accurate and efficient SNN solutions, we emphasize the importance of joint hardware-software codevelopment.

Automated damage assessment of prestressed concrete bridge beam based on sagging

The appropriate bridge maintenance strategy cannot be determined unless the damage is identified, localized, and quantified correctly. Damage assessment can be performed based on model updating, where material properties of a numerical model are modified to represent the damaged state as accurately as possible. However, this approach may become tedious for complex structures such as bridges due to the high number of unknown variables. This study replaces the time-consuming Finite Element (FE) simulations with Artificial Neural Network (ANN) as a surrogate model to reduce the required computational time. The implementation of ANN enables automating the existing manual damage assessment of a prestressed concrete bridge beam. In this paper, the objective is to minimize the difference between the simulated and measured sagging, which is the irreversible downward movement of the bridge due to its weight. The minimization is performed with the Simulated Annealing (SA) algorithm, and the optimization process is repeated with 100 different starting points to ensure robustness. The results indicate that the automated approach performs similarly to the manual approach while being faster and enabling wider exploration in the search space without compromising accuracy. The proposed approach serves as a practical tool for real-world problems by offering an efficient damage assessment.

Implementation of artificial neural network using Levenberg Marquardt algorithm for Casson–Carreau nanofluid flow over exponentially stretching curved surface

Implementation of artificial neural network using Levenberg Marquardt algorithm for Casson–Carreau nanofluid flow over exponentially stretching curved surface

Efficient implementation of the functional links artificial neural networks with cross-terms for nonlinear active noise control

This paper proposes an efficient extension of functional links artificial neural networks (EE-FLANN) for the active noise control (ANC) application. The developed EE-FLANN controller can upgrade the model accuracy with the actual system thanks to adding the cross-terms to the trigonometric function. Unlike the method in the generalized FLANN (GFLANN) controller, the EE-FLANN exploits include cross-term symmetry. However, this causes the computational burden to increase remarkably. To reduce this disadvantage, we truncate the cross-terms appropriately based on the simplified strategy. Furthermore, the adaptive algorithm is designed to partially update the filter coefficients appropriately. Specifically, the cross-terms that do not satisfy certain magnitude conditions will be omitted during the update process to reduce costs. Experiments have shown that the proposed EE-FLANN controller can achieve comparable performance to the GFLANN controller but the complexity is reduced by up to 20%.

Implementation of Artificial Neural Network in Predicting CPO Prices Using Backpropagation

This research examines the development of a forecasting model for Crude Palm Oil (CPO) prices using artificial neural network algorithms, particularly the backpropagation algorithm. CPO, as Indonesia's main export commodity, has significant economic impacts and affects the income of oil palm farmers. Data on CPO prices taken from CIF Rotterdam from January 2019 to December 2023 were used in this study. The research method involved several stages, including data collection, pre-processing, model design, and model implementation using Python programming. The results of training the model using the backpropagation algorithm showed an error value of 0.537829578 after 1000 epochs, while evaluation using Mean Squared Error (MSE) showed an MSE value of 0.022709 during the training process and 0.017604 during the testing process. The model also produced predictions for CPO prices in the next three months: 932.578 for the first month, 949.568 for the second month, and 774.855 for the third month. These findings indicate that the developed model can predict future CPO prices with adequate accuracy, which can assist companies in making better financial decisions and managing risks associated with CPO price fluctuations.

Artificial neural networks trained on simulated multispectral data for real-time imaging of skin microcirculatory blood oxygen saturation.

Imaging blood oxygen saturation ( ) in the skin can be of clinical value when studying ischemic tissue. Emerging multispectral snapshot cameras enable real-time imaging but are limited by slow analysis when using inverse Monte Carlo (MC), the gold standard for analyzing multispectral data. Using artificial neural networks (ANNs) facilitates a significantly faster analysis but requires a large amount of high-quality training data from a wide range of tissue types for a precise estimation of . We aim to develop a framework for training ANNs that estimates in real time from multispectral data with a precision comparable to inverse MC. ANNs are trained using synthetic data from a model that includes MC simulations of light propagation in tissue and hardware characteristics. The model includes physiologically relevant variations in optical properties, unique sensor characteristics, variations in illumination spectrum, and detector noise. This approach enables a rapid way of generating high-quality training data that covers different tissue types and skin pigmentation. The ANN implementation analyzes an image in 0.11s, which is at least 10,000 times faster than inverse MC. The hardware modeling is significantly improved by an in-house calibration of the sensor spectral response. An in-vivo example shows that inverse MC and ANN give almost identical values with a mean absolute deviation of 1.3%-units. ANN can replace inverse MC and enable real-time imaging of microcirculatory in the skin if detailed and precise modeling of both tissue and hardware is used when generating training data.

Phasor-based analysis of a neuromorphic architecture for microwave sensing

This article presents a design procedure for implementing artificial neural networks (ANNs) using conventional microwave components at the hardware level with potential applications in radar and remote sensing. The main objective is to develop structured hardware design methods for implementing artificial neurons, utilizing microwave devices to create neuromorphic devices compatible with high-frequency electromagnetic waves. The research aims to address the challenge of encoding and modulating information in electromagnetic waves into a format suitable for the neuromorphic device by using frequency-modulated information instead of intensity-modulated information. It also proposes a method for integrating principal component analysis as a dimensionality reduction technique with the implementation of ANNs on a single hardware. As a dummy task, the process outlined here is used to implement an artificial neural network at the hardware level, with a specific emphasis on creating hardware that is capable of performing matrix multiplications in the form of dot products while also being able to extract the resulting data in an interpretable manner. The proposed implementation involves the use of directional couplers to implement weights and sample the resulting signal at specific intervals to obtain the multiplication result.

Recent Trends in Application of Memristor in Neuromorphic Computing: A Review

Abstract: Recently memristors have emerged as a form of nonvolatile memory that is based on the principle of ion transport in solid electrolytes under the impact of an external electric field. It is perceived as one of the key elements to building next-generation computing systems owing to its peculiar resistive switching characteristics. The switching mechanism in a memristor is mainly governed by filamentary conduction. Further, it can be employed as a memory as well as a logic element, which makes it an ideal candidate for building innovative computer architecture. Moreover, it is capable of mimicking the characteristics of biological synapses, which makes it an ideal candidate for developing a Neuromorphic system. In this review to begin with the switching mechanism of the memristor, primarily focusing on filamentary conduction, is discussed. Few SPICE models of memristor are reviewed, and their critical comparison is performed, which are widely used to build computing systems. An in-depth study on the various crossbar memory architecture augmented with memristors is reviewed. Finally, the application of memristors in neuromorphic computing and hardware implementation of Artificial Neural Networks (ANN) employing memristors is discussed.

A Study on h‐BN Resistive Switching Temporal Response

AbstractPrevious work that studied hexagonal boron nitride (h‐BN) memristor DC resistive‐switching characteristics is extended to include an experimental understanding of their dynamic behavior upon programming or synaptic weight update. The focus is on the temporal resistive switching response to driving stimulus (programming voltage pulses) effecting conductance updates during training in neural network crossbar implementations. Test arrays are fabricated at the wafer level, enabled by the transfer of CVD‐grown few‐layer (8 layer) or multi‐layer (18 layer) h‐BN films. A comprehensive study of their temporal response under various conditions–voltage pulse amplitude, edge rate (pulse rise/fall times), and temperature–provides new insights into the resistive switching process toward optimized devices and improvements in their implementation of artificial neural networks. The h‐BN memristors can achieve multi‐state operation through ultrafast pulsed switching (< 25 ns) with high energy efficiency (≈10 pJ pulse−1).

Designing machine learning based intelligent network for assessment of heat transfer performance of ternary hybrid nanofluid flow between a cone and a disk: Case of MLP feed forward neural network

In the current study, authors have studied the heat transfer through ternary hybrid nanofluid (THNF) between the gap of a disk and cone, where both are co-rotating with regard to the other. The authors have developed a mathematical model of THNF flow inside a gap between a cone and a disk with the effects of magnetic field, Hall effects, and radiation parameters. The unique aspect of this study is the implementation of a powerful artificial neural network (ANN) that exhibits improved robustness and accurate predictive performance for the Nusselt number at the surface of both the cone and disk. Through the use of proper similarity transformations, the model equations are transformed into a collection of nonlinear ODEs. Using MATLAB and an inbuilt bvp4c function, these equations are numerically solved. The main objective of this research is to examine the impacts of cone and disk rotation within the co-rotation system. The heat transfer rate is seen to be higher at the surface of the disk. It is followed from the analysis of the ANN model that the regression value for each case varies between 0.99574 and 1, and in many cases, it is exactly equal to 1. In addition, the mean square error (MSE) for every dataset is notably very close to zero. The effect of the angle of the conical gap is also one of the noteworthy subjects due to the significance of its effects in several engineering sector applications.

Implementation of Backpropagation Artificial Neural Network for Electricity Load Forecasting in Jember District

The increase in population and various kinds of human activities in the world has made it possible for changes to increase the need for electrical power with demand that is not the same at any time. Based on this description, this research will propose research on the theme of electricity load forecasting as a preventive measure to determine future electricity load needs. Research was assisted using MATLAB data processing software to process research data. Three forecasting models were carried out, namely day, night and day-night conditions. From these three forecasting models, parameters such as epoch, number of input layers, number of hidden layers, activation function, and etc. The data is divided into two parts, training data and test data with a ratio of 70: 30. Test results using the backpropagation artificial neural network method show the highest MSE values for the three forecasting models, day, night, and day-night, are, 0.0039, 0.0041, and 0.002 while the lowest MSE values were in the three models are, 6.77E-04, 0.001, and 0.0011.

Advancing Environmental Sustainability through Implementation of Artificial Neural Networks for Wastewater Treatment Model Prediction and Remediation

Advancing Environmental Sustainability through Implementation of Artificial Neural Networks for Wastewater Treatment Model Prediction and Remediation

Clinical Impact of an AI Decision Support System for Detection of Intracranial Hemorrhage in CT Scans.

This study aimed to evaluate the predictive value and clinical impact of a clinically implemented artificial neural network software model. The software detects intracranial hemorrhage (ICH) from head computed tomography (CT) scans and artificial intelligence (AI)-identified positive cases are then annotated in the work list for early radiologist evaluation. The index test was AI detection by the program Zebra Medical Vision-HealthICH+. Radiologist-confirmed ICH was the reference standard. The study compared whether time benefits from using the AI model led to faster escalation of patient care or surgery within the first 24 h. A total of 2,306 patients were evaluated by the software, and 288 AI-positive cases were included. The AI tool had a positive predictive value of 0.823. There was, however, no significant time reduction when comparing the patients who required escalation of care and those who did not. There was also no significant time reduction in those who required acute surgery compared with those who did not. Among the individual patients with reduced time delay, no cases with evident clinical benefit were identified. Although the clinically implemented AI-based decision support system showed adequate predictive value in identifying ICH, there was no significant clinical benefit for the patients in our setting. While AI-assisted detection of ICH shows great promise from a technical perspective, there remains a need to evaluate the clinical impact and perform external validation across different settings.

Implementation of Exploratory Data Analysis and Artificial Neural Networks to Predict Student Graduation on-Time

Almost all universities in Indonesia face the problem of a low number of students graduating on time. This will affect higher education accreditation. For this reason, universities must pay attention to the timely graduation of their students. The way that can be taken is to predict students' graduation on time. This research aims to predict students' timely graduations using a combination of exploratory data analysis and artificial neural networks. Exploratory data analysis is used to study the relationship between features that influence students' on-time graduation, while artificial neural networks are used to predict on-time graduation. This research goes through method stages, starting with determining the dataset, exploratory data analysis, data preprocessing, dividing training and test data, and applying artificial neural networks. From the research, it was found that Work features and GPS features greatly influence graduation on time. Students who study while working are less likely to graduate on time compared to students who do not work. Students who have an average GPS above 3.00 for eight consecutive semesters will find it easier to graduate on time. Meanwhile, Age and Gender features have no effect on graduating on time. With a percentage of 50% training data and 50% test data, epoch 100, and learning rate 0.001, the best network model was obtained to predict graduation on time with an accuracy rate of 69.84%. The research results also show that the amount of test data and the learning rate can influence the level of accuracy. Meanwhile, the number of epochs does not affect the level of accuracy.

Parametric appraisal and wear prediction of hybrid composites using an integrated soft computing approach

AbstractThis article reports on the implementation of artificial neural network (ANN) and statistical methods to analyze and predict the erosion performance of titania (TiO2) filled ramie‐epoxy based composites. These hybrid composites are prepared by conventional hand lay‐up route and subjected to solid particle erosion tests as per Taguchi's L27 orthogonal array. The effects of different control factors on erosion rate of these composites are studied using Taguchi method. It reveals that impact velocity and filler content have significant effect on the erosion wear rate followed by other least significant factors. The individual effect of each control factor while keeping other factors constant is ascertained by performing steady state erosion experiments. Further, a computational model based on ANN is used as a tool to effectively predict the erosion rates of the composites. The results show that the predicted values are in reasonably good agreement with the experimental ones with an accuracy of 90% and relative error lying within a range of 1%–10%. Further, the trained ANN model is validated by considering the erosion rates obtained during steady state erosion process as the input parameter. The possible mechanisms causing the wear loss are identified using electron microscopy.Highlights Successful fabrication of titanium oxide reinforced hybrid composites. Steady state erosion experiments are performed. Erosion rates are predicted using ANN and compared with the experimental data. Wear loss mechanisms are identified using electron microscopy.