Neural Network Learning Algorithm Research Articles

Estimation of body weight from morphological measurements in Boer goats with the application of artificial neural networks and some regression models

In this study, examination of the characteristics of body measurements affecting the body weight of Boer goats and the estimation of the body weight were investigated. To examine their body morphological features, 400 live animals were taken into consideration. The morphological measurements taken from the goats in the study were body weight (BW), body length (BL), heart girth (HG), withers height (WH), rump height (RH), rump length (RL), ear length (EL) and head with (HW) respectively. These animals were between 1-6 years old; 112 of them were male and 288 of them were female. Multiple regression, ridge regression and artificial neural networks (ANN) methods were applied to estimate the body weight. In the prediction of body weight as a dependent variable, the ANNs predictive model produced high predictive performance. Mean square error (MSE), mean absolute error (MAD) and mean absolute percent error (MAPE) statistics were used to determine model performance. Using the Multi-Layer Perceptron (MLP) Artificial Neural Network (ANN) learning algorithm, the body features that had the greatest impact on body weight were determined. Comparison of the predictive performance of the put forward model against both multiple regression and state of the ridge regression methods showed that the artificial neural networks outperformed both competing models by achieving the least values for MAD, MSE and MAPE in both training and testing data sets. The results of artificial neural networks were promising and accurate in the prediction of the body weight of goats.

Online course resource recommendation based on deep learning

In order to further improve the online learning experience and effect of courses for learners, this study, with the help of deep learning neural network algorithm, based on the characteristics of deep learning algorithm such as good processing performance and strong data adaptability, carries out research for online course students' preferences, learner characteristics and other fields, and constructs a course resource recommendation technical algorithm based on deep learning. Through algorithm experiments, it is found that the performance of NCF model is improved by 57.7% compared with that of CF model in the algorithm with RMSE as evaluation index. The performance under MAE evaluation index is improved by 58.5%, indicating that the integration of the deep learning neural network algorithm can better improve the online course recommendation effect and enhance students' online learning experience.

Small-signal Modeling of InP HBT based on PSO-ELM Neural Network

Extreme Learning Machine (ELM) is a new learning algorithm for single-hidden layer feedforward neural network, which has been widely used in lots of fields. However, it still has the insufficiency of randomly determining the hidden layer threshold and output weight, which leads to ill-conditioned output. In order to avoid the risk of decreasing prediction accuracy caused by this possibility, the ELM is optimized using particle swarm algorithm. A Particle Swarm Optimization (PSO) enhanced ELM algorithm is proposed to accurately model the small-signal properties of InP Heterojunction Bipolar Transistors (HBTs). PSOELM algorithm solves the problem of unstable prediction data caused by random determination of input weights in ELM. Comparing the modeling effects of the PSO-ELM model and the ELM model under different bias conditions for a 1 μm×15 μm InP HBT, it is proved that the PSO-ELM algorithm has better consistency with the measured data.

Statistical Analysis with Machine and Neural Learning-Based Model on Cardiovascular Diseases and Stroke Prediction

Several risk factors, such as hypertension, hyperlipidemia, and an irregular heart rhythm, make an early diagnosis of cardiovascular disease challenging. Reducing cardiac risk calls for precise diagnosis and therapy. Clinical practice in the healthcare business is likely to evolve in tandem as a result of advancements in machine learning. Therefore, scientists and doctors need to acknowledge machine learning's significance. The fundamental purpose of this research is to a reliable analyzing Risk Factors for Cardiovascular Disease method that makes use of machine learning. Classifying well-known cardiovascular datasets But, on the other hand, is a job for state-of-the-art machine learning techniques and neural network algorithms. Several statistical and visualization indicators were used to assess the efficacy of the suggested approaches and to determine the optimal machine-learning and neural-network approach. Using these modeling methods acquired high and accurate accuracy on stroke and heart disease prediction.

Comparative Analysis of Deep Learning Models on Brain Tumor Segmentation Datasets: BraTS 2015-2020 Datasets

Deep Learning neural networks have shown applicability in segmentation of brain tumor images.This research have been carried for comprehensive review of several deep learning neural networks. The datasets included in this study are standard datasets Multimodal Brain Tumor Segmentation (BraTS). This paper has summarized the performance of various deep learning neural network algorithms on BraTS datasets. Algorithms have been compared and summarized against the baseline models with specific attributes like dice score, PPV and sensitivity. It has been found that out of the different models applied on the BraTS 2015 dataset GAN in the year 2020 algorithm is showing better results on this data set. GAN architecture termed RescueNet gave the best segmentation results in terms of 0.94 dice score and 0.88 Sensitivity. This has been also observed that models used cascaded deep learning models had independent deep learning models at each stage which had no correlation among the stages which can cause class imbalance. Further it have found that the Attention models tried to solve problem of class imbalance in the brain tumor segmentation task. This work also found that existing CNN’s is having overfitting issues. For this ResNet models can add a rapid connect bounce relationship parallel to the layers of CNN to accomplish better outcomes for the brain tumor segmentation task.

SYMBIOSIS OF MEDICAL TECHNOLOGIES AND ARTIFICIAL INTELLIGENCE: NEW OPPORTUNITIES IN ONCOLOGY

SYMBIOSIS OF MEDICAL TECHNOLOGIES AND ARTIFICIAL INTELLIGENCE: NEW OPPORTUNITIES IN ONCOLOGY

An Enhanced LBPH Approach to Ambient-Light-Affected Face Recognition Data in Sensor Network

Although combining a high-resolution camera with a wireless sensing network is effective for interpreting different signals for image presentation on the identification of face recognition, its accuracy is still severely restricted. Removing the unfavorable impact of ambient light remains one of the most difficult challenges in facial recognition. Therefore, it is important to find an algorithm that can capture the major features of the object when there are ambient light changes. In this study, face recognition is used as an example of image recognition to analyze the differences between Local Binary Patterns Histograms (LBPH) and OpenFace deep learning neural network algorithms and compare the accuracy and error rates of face recognition in different environmental lighting. According to the prediction results of 13 images based on grouping statistics, the accuracy rate of face recognition of LBPH is higher than that of OpenFace in scenes with changes in ambient lighting. When the azimuth angle of the light source is more than +/−25° and the elevation angle is +000°, the accuracy rate of face recognition is low. When the azimuth angle is between +25° and −25° and the elevation angle is +000°, the accuracy rate of face recognition is higher. Through the experimental design, the results show that, concerning the uncertainty of illumination angles of lighting source, the LBPH algorithm has a higher accuracy in face recognition.

FNNS: An Effective Feedforward Neural Network Scheme with Random Weights for Processing Large-Scale Datasets

The size of datasets is growing exponentially as information technology advances, and it is becoming more and more crucial to provide efficient learning algorithms for neural networks to handle massive amounts of data. Due to their potential for handling huge datasets, feed-forward neural networks with random weights (FNNRWs) have drawn a lot of attention. In this paper, we introduced an efficient feed-forward neural network scheme (FNNS) for processing massive datasets with random weights. The FNNS divides large-scale data into subsets of the same size, and each subset derives the corresponding submodel. According to the activation function, the optimal range of input weights and biases is calculated. The input weight and biases are randomly generated in this range, and the iterative scheme is used to evaluate the output weight. The MNIST dataset was used as the basis for experiments. The experimental results demonstrate that the algorithm has a promising future in processing massive datasets.

Supervised Learning in Neural Networks: Feedback-Network-Free Implementation and Biological Plausibility.

The well-known backpropagation learning algorithm is probably the most popular learning algorithm in artificial neural networks. It has been widely used in various applications of deep learning. The backpropagation algorithm requires a separate feedback network to back propagate errors. This feedback network must have the same topology and connection strengths (weights) as the feed-forward network. In this article, we propose a new learning algorithm that is mathematically equivalent to the backpropagation algorithm but does not require a feedback network. The elimination of the feedback network makes the implementation of the new algorithm much simpler. The elimination of the feedback network also significantly increases biological plausibility for biological neural networks to learn using the new algorithm by means of some retrograde regulatory mechanisms that may exist in neurons. This new algorithm also eliminates the need for two-phase adaptation (feed-forward phase and feedback phase). Hence, neurons can adapt asynchronously and concurrently in a way analogous to that of biological neurons.

Gender recognition based fingerprints using dynamic horizontal voting ensemble deep learning

Despite tremendous advancements in gender equality, there are still persistent gender disparities, especially in important human activities. Consequently, gender inequality and related concerns are serious problems in our global society. Major players in the global economy have identified the gender identity system as a crucial stepping stone for bridging the enormous gap in gender-based problems. Extensive research conducted by forensic scientists has uncovered a unique pattern in the fingerprint, and these distinguishing characteristics of fingerprints can be utilized to determine the gender of individuals. Numerous research has revealed various fingerprint-based approaches to gender recognition. This research aims to present a novel dynamic horizontal voting ensemble model with a hybrid Convolutional Neural Network and Long Short Term Memory (CNN-LSTM) deep learning algorithm as the base learner to determine human gender attributes based on fingerprint patterns automatically. More than four thousand Live fingerprint images were acquired and subjected to training, testing, and classification using the proposed model. The results of this study indicated over 99% accuracy in predicting a person’s gender. The proposed model also performed better than other state-of-the-art models, such as ResNet-34, VGG-19, ResNet-50, and EfficientNet-B3, when implemented on the SOCOFing public dataset.

An approximate randomization-based neural network with dedicated digital architecture for energy-constrained devices

Variable energy constraints affect the implementations of neural networks on battery-operated embedded systems. This paper describes a learning algorithm for randomization-based neural networks with hard-limit activation functions. The approach adopts a novel cost function that balances accuracy and network complexity during training. From an energy-specific perspective, the new learning strategy allows to adjust, dynamically and in real time, the number of operations during the network’s forward phase. The proposed learning scheme leads to efficient predictors supported by digital architectures. The resulting digital architecture can switch to approximate computing at run time, in compliance with the available energy budget. Experiments on 10 real-world prediction testbeds confirmed the effectiveness of the learning scheme. Additional tests on limited-resource devices supported the implementation efficiency of the overall design approach.

IMPLEMENTASI ALGORITMA BACKPROPAGATION UNTUK MEMPREDIKSI JUMLAH JIWA TERDAMPAK BANJIR DI DKI JAKARTA

<p class="MsoNormal" style="margin-bottom: .0001pt; text-align: justify; line-height: normal; mso-layout-grid-align: none; text-autospace: none;"><span style="mso-bidi-font-size: 12.0pt; font-family: 'Times New Roman',serif; color: black;" lang="EN-US">Floods are natural disasters that often occur in the DKI Jakarta area. DKI Jakarta government needs to anticipate the impact of the flood disaster by estimating the number of people affected by the flood. The number of people affected by floods that are uncertain every month can be predicted for the future. There are many ways that can be predict the number of people affected by floods, one of them is artificial neural network method. One of learning algorithms in artificial neural networks is backpropagation algorithm. This research applies an artificial neural network method with backpropagation algorithm to predict the number of people affected by floods in DKI Jakarta. In this research, training process was carried out 100 times on each network architecture (12-10-1, 12-12-1, 12-14-1) with several parameters such as epoch, momentum constant, and learning rate. The best results in the training process are carried out to testing process to test the network. In the testing process, the best results are 12-10-1 architecture with an accuracy rate 98.704%. Based on these results, it can be said that this network can predict well and can be implemented for forecasting the number of people affected by floods in DKI Jakarta.</span></p>

CNN-LSTM deep learning based forecasting model for COVID-19 infection cases in Nigeria, South Africa and Botswana.

COVID-19 pandemic has indeed plunged the global community especially African countries into an alarming difficult situationculminating into a great deal amounts of catastrophes such as economic recession, political instability and loss of jobs. Thepandemic spreads exponentially and causes loss of lives. Following the outbreak of the omicron new variant of concern,forecasting and identification of the COVID-19 infection cases is very vital for government at various levels. Hence, havingknowledge of the spread at a particular point in time, swift actions can be taken by government at various levels with a view toaccordingly formulate new policies and modalities towards minimizing the trajectory of the consequences of COVID-19 pandemicto both public health and economic sectors. Here, a potent combination of Convolutional Neural Network (CNN) learning algorithm along with Long Short Term Memory(LSTM) learning algorithm has been proposed in this work in order to produce a hybrid of a deep learning algorithm ConvolutionalNeural Network - Long Short Term Memory (CNN-LSTM) for forecasting COVID-19 infection cases particularly in Nigeria, SouthAfrica and Botswana. Forecasting models for COVID-19 infection cases in Nigeria, South Africa and Botswana, were developedfor 10 days using deep learning-based approaches namely CNN, LSTM and CNN-LSTM deep learning algorithm respectively. The models were evaluated on the basis of four standard performance evaluation metrics which include accuracy, MSE, MAEand RMSE respectively. However, the CNN-LSTM deep learning-based forecasting model achieved the best accuracy of98.30%, 97.60%, and 97.74% for Nigeria, South Africa and Botswana respectively; and in the same manner, achieved lesserMSE, MAE and RMSE values compared to models developed with CNN and LSTM respectively. Taken together, the CNN-LSTM deep learning-based forecasting model for COVID-19 infection cases in Nigeria, South Africaand Botswana dramatically surpasses the two other DL based forecasting models (CNN and LSTM) for COVID-19 infectioncases in Nigeria, South Africa and Botswana in terms of not only the best accuracy of with 98.30%, 97.60%, and 97.74% but alsoin terms of lesser MSE, MAE and RMSE.

Adaptive Generalized Eigenvector Estimating Algorithm for Hermitian Matrix Pencil

Generalized eigenvector plays an essential role in the signal processing field. In this paper, we present a novel neural network learning algorithm for estimating the generalized eigenvector of a Hermitian matrix pencil. Differently from some traditional algorithms, which need to select the proper values of learning rates before using, the proposed algorithm does not need a learning rate and is very suitable for real applications. Through analyzing all of the equilibrium points, it is proven that if and only if the weight vector of the neural network is equal to the generalized eigenvector corresponding to the largest generalized eigenvalue of a Hermitian matrix pencil, the proposed algorithm reaches to convergence status. By using the deterministic discretetime (DDT) method, some convergence conditions, which can be satisfied with probability 1, are also obtained to guarantee its convergence. Simulation results show that the proposed algorithm has a fast convergence speed and good numerical stability. The real application demonstrates its effectiveness in tracking the optimal vector of beamforming.

Prediction of Ground Surface Deformation Induced by Earthquake on Urban Area Using Machine Learning

Earthquakes can inflict significant damage to structures and infrastructures. This paper presents a machine learning model to predict ground surface deformation (GDS) induced by earthquake events. The data on historical GSD is extracted from radar product of Synthetic Aperture Radar (SAR) data of one-year over five magnitude earthquakes that occurred within 200 kilometers of the Kota Padang Regency, West Sumatra. Building topology data of its footprint area, distance from shoreline, elevation, and coordinate were incorporated as the main features in the dataset. The earthquake parameters were taken from the USGS earthquake data catalog. Four machine learning algorithms of Neural Network (NN), Random Forest (RF), k-Nearest Neighbors (kNN), and Gradient Boosting (GB) are applied. The GSD from the trained models is predicted and compared with the measured GSD from the SAR’s product. The performances of proposed algorithms are evaluated in terms of the statistical index. A new dataset from the earthquake event in March 2022 is used to predict the GSD and further test the performance of the trained models. Overall, the four machine learning algorithms have outstanding performance, with a coefficient determinant of more than 0.9. The kNN algorithm outperforms compared to others in delineating the GSD. The trained models gave deficient prediction performance on the new dataset with a correlation coefficient of 0.228 predicted by the RF algorithm. Additional earthquake datasets and more unique features will improve the performance of the machine learning algorithms.

Identification of the Angle Errors of the LED Parallel-Light Module in PCB Exposure Device by Using Neural Network Learning Algorithms

For the smart manufacturing development of printed-circuit-board (PCB) exposure devices, the LED parallel-light (LPL) module is investigated and the angle errors of those LPL units are identified by neural network learning algorithms. At present, in PCB manufacturing, most circuit boards use photoresist covering etching. After exposure and development, unwanted copper foil is etched and removed to make circuit boards. The exposure process is its key process, and the equipment used in this process is an exposure machine. The LPL unit is designed and the LPL exposure module is searched under the principle of higher irradiance uniformity. The learning data of supervised learning for the convolutional neural network (CNN) include a 2D irradiance distribution image constructed by the ray tracing simulation tool. In these supervised learning data, all units of LPL-EM are randomly added with a self-specific angle error. By using Fast Region-based CNN, the identification of the multi-LPL module with the specific errors of inclination and azimuth angle is verified. Those results preliminarily illustrate that supervised learning techniques should be able to help identify the errors of inclination and azimuth angle for the single LPL unit and multi-light module of PCB exposure devices. In other words, this technology should serve as a reference for the development of the PCB exposure process towards smart manufacturing.

Multiple learning neural network algorithm for parameter estimation of proton exchange membrane fuel cell models

Extracting the unknown parameters of proton exchange membrane fuel cell (PEMFC) models accurately is vital to design, control, and simulate the actual PEMFC. In order to extract the unknown parameters of PEMFC models precisely, this work presents an improved version of neural network algorithm (NNA), namely the multiple learning neural network algorithm (MLNNA). In MLNNA, six learning strategies are designed based on the created local elite archive and global elite archive to balance exploration and exploitation of MLNNA. To evaluate the performance of MLNNA, MLNNA is first employed to solve the well-known CEC 2015 test suite. Experimental results demonstrate that MLNNA outperforms NNA on most test functions. Then, MLNNA is used to extract the parameters of two PEMFC models including the BCS 500 W PEMFC model and the NedStack SP6 PEMFC model. Experimental results support the superiority of MLNNA in the parameter estimation of PEMFC models by comparing it with 10 powerful optimization algorithms.

High-performance multi-parameter fiber sensor by grating-enhanced Mach-Zehnder interference.

A multi-parameter dual-core fiber sensor is proposed to realize highly sensitive detection of illumination, temperature, and humidity, separately. Through partial grating etching of a one-side core, the interaction between the core and the external environment is enhanced. Then, combining the Mach-Zehnder effect of the dual core, a higher sensing sensitivity is obtained. Experimental results show the temperature sensitivity is higher than 6.1952 nm/°C. Besides, the humidity and illumination resolution can reach as accurate as 0.041 relative humidity (RH) and 0.025 light units, respectively. To have better multi-parameter sensing and demodulation, the deep learning algorithm of a one-dimensional convolutional neural network (1D-CNN) is used to reach an accuracy of 99.05% with ∼2.00 root mean square error (RMSE). We envision such an excellent multi-parameter sensor can be promising in environmental monitoring and intelligent manufacturing.

NcRNAInter: a novel strategy based on graph neural network to discover interactions between lncRNA and miRNA.

In recent years, many studies have illustrated the significant role that non-coding RNA (ncRNA) plays in biological activities, in which lncRNA, miRNA and especially their interactions have been proved to affect many biological processes. Some in silico methods have been proposed and applied to identify novel lncRNA-miRNA interactions (LMIs), but there are still imperfections in their RNA representation and information extraction approaches, which imply there is still room for further improving their performances. Meanwhile, only a few of them are accessible at present, which limits their practical applications. The construction of a new tool for LMI prediction is thus imperative for the better understanding of their relevant biological mechanisms. This study proposed a novel method, ncRNAInter, for LMI prediction. A comprehensive strategy for RNA representation and an optimized deep learning algorithm of graph neural network were utilized in this study. ncRNAInter was robust and showed better performance of 26.7% higher Matthews correlation coefficient than existing reputable methods for human LMI prediction. In addition, ncRNAInter proved its universal applicability in dealing with LMIs from various species and successfully identified novel LMIs associated with various diseases, which further verified its effectiveness and usability. All source code and datasets are freely available at https://github.com/idrblab/ncRNAInter.

THE FRACTAL AND PIECEWISE STRUCTURE OF SOME CHAOTIC NEURAL NETWORKS USING A GENERALIZED MODEL

Fractal structures are everywhere around us as they occur naturally or are artificially simulated. Applied mostly in engineering domains that include neural networks, fractal processes help fostering architectural design. For instance, fractal models are commonly used to design new machine learning algorithms for neural networks. Differential operators that can artificially trigger such fractal processes become a valuable asset for engineers. We use in this paper the fractal derivative combined to the fractional dynamic to analyze the chaotic proto-Lü system. That combined operator, known as the fractal-fractional derivative (FFD), is relatively new in the literature and has many features still to be discovered. The piecewise model of the proto-Lü system combining the fractal-fractional and classical derivatives is analyzed and solved numerically. In the study, we start by providing a succinct summary of fundamentals behind the FFD and equations of the proto-Lü system. The latter comprise different models with [Formula: see text] scrolls each ([Formula: see text]) corresponding to its nth cover. We focus on applying the FFD on the third cover of the proto-Lü system that is solved numerically via the Haar wavelets. Numerical simulations show the maintenance of the multiscroll chaotic attractors. The representations for the piecewise model also show the maintenance of the triple cover both in a stretched form and a self-similarity process. Additionally, we observe the capacity of those attractors to perform self-replication in a fractal structure that varies with the fractional-order derivative.