Number Of Hidden Neurons Research Articles

Adaptive multilevel attention deeplabv3+ with heuristic based frame work for semantic segmentation of aerial images using improved golden jackal optimization algorithm

Semantic segmentation of small-scale objects in aerial images is challenging due to low-level characteristics in neural networks and varying information in feature maps. Deeplab models struggle with poor edge refinement, resulting in rough borders, and fail to fully exploit relationships between pixel categories at different distances. To addressing the issue in deeplab series, an adaptive multi-level attention based deeplabv3+ (AMLA-Deeplabv3+) with improved golden jackal optimization algorithm is implemented in this paper. Multi-level attention unit has been included in the Atrous spatial pyramid pooling module in the encoder section of deeplabv3+ to bridge the semantic feature gap among encoders output. To put more weights on relevant features squeeze and excitation units has been included in the decoder section of deeplabv3+. The improved golden jackal optimization (IGJO) algorithm is deployed to fine tune the hyper parameters such as number of hidden neurons, epochs, learning rate and batch size of deeplabv3+. The proposed AMLA-Deeplabv3+ is tested on two publicly available datasets named semantic segmentation of aerial imagery and aerial image segmentation. The proposed model achieves better segmentation outcomes in terms of accuracy of 99.65%, Precision of 97.21%, F1-score of 99.48% and mean intersection over union of 98.44% with a computation time of 3.0231 minutes. The experimental outcomes shows that the proposed model’s efficiency is superior to other conventional techniques.

Thermal coal futures trading volume predictions through the neural network

Purpose Predicting commodity futures trading volumes represents an important matter to policymakers and a wide spectrum of market participants. The purpose of this study is to concentrate on the energy sector and explore the trading volume prediction issue for the thermal coal futures traded in Zhengzhou Commodity Exchange in China with daily data spanning January 2016–December 2020. Design/methodology/approach The nonlinear autoregressive neural network is adopted for this purpose and prediction performance is examined based upon a variety of settings over algorithms for model estimations, numbers of hidden neurons and delays and ratios for splitting the trading volume series into training, validation and testing phases. Findings A relatively simple model setting is arrived at that leads to predictions of good accuracy and stabilities and maintains small prediction errors up to the 99.273th quantile of the observed trading volume. Originality/value The results could, on one hand, serve as standalone technical trading volume predictions. They could, on the other hand, be combined with different (fundamental) prediction results for forming perspectives of trading trends and carrying out policy analysis.

Efficient grid management: smart forecasting of short-term power load using PSO-LSTM

Recent load forecasting techniques combining machine learning models and hyperparameter optimization algorithms have shown success for short-term load forecasting (STLF) task, but they often require complex programming, higher computational costs, and greater parameter tuning. In this paper, we introduce an improved STLF model that combines Long Short-Term Memory (LSTM) neural network with Particle Swarm Optimization (PSO) for enhanced performance. In the proposed approach, the number of hidden neurons in different LSTM layers, learning rate and the number of iterations for training are optimized using the PSO algorithm. To validate the effectiveness of this method, meteorological data and historical load data from a real-world power grid are used as input. The experimental results reveal PSO significantly enhances hyperparameter tuning for LSTM neural networks, leading to improved predictive modelling. The PSO-LSTM model performed better than the LSTM model by more than 20% (in terms of Mean Absolute Error), and showed low sensitivity to hyperparameters. Comparative analysis with alternative approaches from the literature further validates the PSO-LSTM’s effectiveness in STLF. Additionally, the model achieved stable multi-step prediction capabilities, with average errors of 3.6445 for MAE, 4.6509 for RMSE, and 4.6519 for MAPE over a 1–4 day ahead lead times. This study highlights PSO-LSTM’s enhanced robustness and accuracy in power load prediction while addressing hyperparameter tuning challenges through self-optimization.

An optimized hybrid finite element analyses - Artificial neural networks technique for estimating in-plane orthotropic mechanical properties of printed circuit boards

Explicitly modeling Printed Circuit Boards (PCBs) in Finite Element Analysis (FEA) is not practical due to the complex multi-component structure of PCBs. To overcome this complication, a PCB can be simplified and modeled as an orthotropic plate with equivalent mechanical properties that result in the same natural frequencies as the actual PCB. This research introduces an optimized hybrid technique combining FEA and Artificial Neural Networks (ANNs) to accurately estimate the equivalent in-plane orthotropic mechanical properties of PCBs. This study presents a systematic approach where natural frequencies obtained from FEA are used to train ANNs for predicting the equivalent representative mechanical properties. The research employs a rigorous optimization procedure involving various ANN configurations (i.e., different learning algorithms, activation functions, and number of hidden neurons) and training dataset sizes. To further ensure the reliability of the results, 10 cross-validation are carried out for each ANN configuration by employing dynamic data-splitting strategy, and all measures used for assessing the accuracy of the ANN predictions are based on averaged results for the 10 cross-validations. The accuracy of the ANN predictions is tested against both simulated and experimental data. The Mean Absolute Percentage Error (MAPE) is found to be about 4 % based on material properties against the simulated data, and about 1.2 % based on natural frequencies against the experimental data point. The results demonstrate superior predictive accuracy and efficiency compared to existing models, highlighting the potential of the proposed approach in advancing the reliability and performance of PCB mechanical property estimations.

Investigating Feed-Forward Back-Propagation Neural Network with Different Hyperparameters for Inverse Kinematics of a 2-DoF Robotic Manipulator: A Comparative Study

Inverse kinematics is a significant challenge in robotic manipulators, and finding practical solutions plays a crucial role in achieving precise control. This paper presents a study on solving inverse kinematics problems using the Feed-Forward Back-Propagation Neural Network (FFBP-NN) and examines its performance with different hyperparameters. By utilizing the FFBP-NN, our primary objective is to ascertain the joint angles required to attain precise Cartesian coordinates for the end-effector of the manipulator. To accomplish this, we first formed three input-output datasets (a fixed-step-size dataset, a random-step-size dataset, and a sinusoidal-signal-based dataset) of joint positions and their respective Cartesian coordinates using direct geometrical formulations of a two-degree-of-freedom (2-DoF) manipulator. Thereafter, we train the FFBP-NN with the generated datasets using the MATLAB Neural Network Toolbox and investigate its potential by altering the hyperparameters (e.g., number of hidden neurons, number of hidden layers, and training optimizer). Three different training optimizers are considered, namely the Levenberg-Marquardt (LM) algorithm, the Bayesian Regularization (BR) algorithm, and the Scaled Conjugate Gradient (SCG) algorithm. The Mean Squared Error is used as the main performance metric to evaluate the training accuracy of the FFBP-NN. The comparative outcomes offer valuable insights into the capabilities of various network architectures in addressing inverse kinematics challenges. Therefore, this study explores the application of the FFBP-NNs in tackling the inverse kinematics, and facilitating the choice of the most appropriate network design by achieving a portfolio of various experimental results by considering and varying different hyperparameters of the FFBP-NN.

Design of Software Reliability Prediction using Radial Basis Function Networks with Nonlinear Analysis and Topological Considerations

This study presents a thorough methodology for predicting software reliability by employing Radial Basis Function Networks (RBFNs). By using machine learning methodologies such as Radial Basis Function Networks (RBFNs), software development teams are enabled to make well-informed decisions, optimize resource allocation, and proactively mitigate potential dependability concerns. Consequently, this results in improved software quality and heightened customer happiness. This methodology involves multiple stages, starting with data collection and preprocessing. We assemble a comprehensive dataset comprising historical software performance metrics, defect reports, and relevant development process information. After data cleansing and feature engineering, we split the dataset into training, validation, and testing subsets. The core of this approach lies in the construction and training of RBFNs. These neural networks consist of an input layer, a hidden layer with radial basis functions, and an output layer. The architecture parameters, particularly the number of hidden neurons and the spread parameter of the radial basis functions, are optimized through a systematic hyperparameter tuning process using the validation dataset. Upon achieving an optimal model configuration, we rigorously evaluate the RBFN predictive capabilities using the testing dataset.

Optimized extreme learning machine using genetic algorithm for short-term wind power prediction

Through the much defiance facing energy today, it has become necessary to rely on wind energy as a source of unlimited renewable energies. However, energy planning and regulation require wind capacity forecasting, because oscillations of wind speed drastically affect directly power generation. Therefore, several scenarios must be provided to allow for estimating uncertainties. To deal with this problem, this paper exploits the major advantages of the regularized extreme learning machine algorithm (R-ELM) and thus proposes a model for predicting the wind energy generated for the next hour based on the time series of wind speed. The R-ELM is combined with the genetic algorithm which is designed to optimize the most important hyperparameter which is the number of hidden neurons. Thus, the proposed model aims to forecast the average wind power per hour based on the wind speed of the previous hours. The results obtained showed that the proposed method is much better than those reported in the literature concerning the precision of the prediction and the time convergence.

Exact, Fast and Expressive Poisson Point Processes via Squared Neural Families

We introduce squared neural Poisson point processes (SNEPPPs) by parameterising the intensity function by the squared norm of a two layer neural network. When the hidden layer is fixed and the second layer has a single neuron, our approach resembles previous uses of squared Gaussian process or kernel methods, but allowing the hidden layer to be learnt allows for additional flexibility. In many cases of interest, the integrated intensity function admits a closed form and can be computed in quadratic time in the number of hidden neurons. We enumerate a far more extensive number of such cases than has previously been discussed. Our approach is more memory and time efficient than naive implementations of squared or exponentiated kernel methods or Gaussian processes. Maximum likelihood and maximum a posteriori estimates in a reparameterisation of the final layer of the intensity function can be obtained by solving a (strongly) convex optimisation problem using projected gradient descent. We demonstrate SNEPPPs on real, and synthetic benchmarks, and provide a software implementation.

Platinum and palladium price forecasting through neural networks

To many commodity market participants, forecasts of price series represent a critical task. In this work, nonlinear autoregressive neural network models’ potential is explored for forecasting daily prices series of platinum and palladium over about a fifty-year period. For this purpose, one hundred and twenty model settings are examined, including different training algorithms, numbers of hidden neurons and delays, and ratios used to segment the data. With the analysis, two models leading to stable and accurate forecast results are constructed for the prices of the two commodities. In particular, the models’ performance in terms of the relative root mean square error is 1.86% and 3.61% for platinum and palladium, respectively, for the overall sample. Results in this work could help technical forecasts and policy analysis. The forecast framework might be extended to other different commodities.

Optimizing Degradable Plastic Density Prediction: A Coarse-to-Fine Deep Neural Network Approach

Density is an important property for the production of high-quality degradable plastics. Density is useful to determine the type of plastic material and to detect physical changes in the plastic material. In this paper, a novel technique for predicting the density of degradable plastics using Deep Neural Networks (DNN) is presented. The aim was to reduce the dimension of the inputs in order to establish a strong relationship between the inputs using principal component analysis (PCA). The results show that the combination of polyethylene, oil palm biomass, starch and palm oil has a greater impact on predicting the density of degradable plastics. Subsequently, the number of hidden neurons is determined by a coarse-to-fine search to develop the network topology of the DNN model for predicting the density of degradable plastics. The developed DNN model consists of 4 input neurons, 62 neurons in the first hidden layer, 31 neurons in the second hidden layer and one output neuron. The developed DNN model showed high accuracy with the lowest values for RMSE, MAE and MSE, indicating that the use of a DNN model is a suitable method for predicting the density of degradable plastics. Furthermore, this study has the potential to make rapid and accurate predictions about the physical properties of degradable plastics in the context of polymers.

Coastal groundwater quality prediction using objective-weighted WQI and machine learning approach.

The water quality index (WQI) is a globally accepted guideline to indicate the water quality standard of any groundwater resource. Water levels in existing groundwater sources are declining in several coastal zones. Therefore, for monitoring water quality and improving water management, the prediction and identification of groundwater status by an effective technique with higher accuracy is urgently needed. Therefore, this research aims to find an effective model for WQI prediction by comparing entropy and critic weight-based WQI (ENW-WQI and CRITIC-WQI) with multi-layer perceptron artificial neural network (MLP-ANN) technique and also to identify contaminated zones using GIS. Initially, 1000 water sampling datasets with concentrations of several water quality parameters of different coastal blocks of eastern India during 2018 to 2022 are considered for the estimation of ENW-WQI and CRITIC-WQI. It shows 65% and 67% of the samples are excellent to good for drinking. ENW-WQI and CRITIC-WQI-based MLP-ANN models have been established considering different data portioning and hidden neuron numbers. Input variables and appropriate dataset partitioning with hidden neurons for models obtained from correlation and trial-error analysis. Spatial distribution maps are also produced for calculated WQIs using inverse distance weighted interpolation approaches. Three fitting models are obtained: ENW-WQI-MLP-ANN, CRITIC-WQI-MLP-ANN-I and CRITIC-WQI-MLP-ANN-II. CRITIC-WQI-MLP-ANN-II model (data ratio 85:15, network structure 6-12-1, R2 = 0.986, NSE = 0.98, and error rate 0.49%) provides the best accuracy in WQI prediction. The GIS-based WQI maps record several areas related to drinking water quality. The results of this research can help in planning the provision of safe drinking water in the future.

PQD classifier based on higher-order statistics and total harmonic distortion

Higher-Order Statistics (HOS) have been frequently applied in Power Quality Disturbance (PQD) analysis as a reliable tool for event detection. This paper outlines a technique based on mean, variance and zero-lag third and fourth cumulants – skewness and kurtosis – along with the Total Harmonic Distortion (THD) index for PQD detection. These statistics are obtained in order to characterize a waveform by a feature vector. A two-layer feed-forward neural network is then used to classify inputs (feature vectors) into a set of PQD categories. The impact of frame duration and number of hidden neurons is analyzed. The network is trained, validated and tested with synthetically-generated PQD waveforms obtained from parameter-controlled equations. As a first approach, five PQD categories are considered: sag, swell, interruption, impulsive transient and oscillatory transient. A promising overall classification rate of 99.7 % is achieved which allows future analysis with more PQD categories and/or a noisy context.

Research the adaptive neural networks using possibility in control systems for cycle chemistry at thermal power plants

The study is devoted to the analysis of the learning possibilities of adaptive neural networks in the context of water-chemical regime management systems. During a series of experiments on a laboratory stand with a disturbing effect, transient characteristics describing changes in the controlled value were obtained. The obtained mathematical models using the TunePID program served as the basis for the analysis and research of adaptive neural networks. Adaptive neural networks are a powerful tool capable of learning from the data obtained and dynamically changing their structure and parameters to accurately predict and control the behavior of the system. Unlike classical mathematical models, adaptive neural networks have the ability to process nonlinear dependencies and adapt to changing conditions without the need for manual reconfiguration. The stages of identification of disturbances and adaptation of compensator settings were described and illustrated. The study included an analysis of the work of a neural network with a different number of hidden neurons and various activation functions. The results showed that neural networks trained on data using an adaptive approach are able to predict and control the system with an accuracy comparable to the calculated parameters of the control coefficients calculated manually.

Passenger Flow Prediction and Risk Early Warning for Metro Cross-section based on Transportation Big Data

With the increasing pace of urbanization and the diversification of urban transportation modes, the demand for urban transportation in people's life is also increasing. For the purpose of relieving urban traffic, cities have made great efforts to develop public transport, especially metro transport. This paper will use the subway swipe card data of Changsha city from March 1-15, 2023, and optimize the LSTM model based on particle swarm algorithm, to predict the metro cross-sectional passenger volume. Based on the data and methods, this paper investigates the short-time cross-sectional passenger flow forecasting of the metro using the variables of initial passenger flow, inbound passenger volume and outbound passenger volume as references. It is found that the combined short-time passenger flow prediction model (PSO-LSTM) has good prediction effect. Based on the standard LSTM model, the PSO algorithm optimizes the learning rate, the number of iterations and the number of hidden neurons in the LSTM parameters to construct the PSO-LSTM model. The video recognition technology more accurately reflects the traffic congestion and makes early warning. The short-time cross-sectional passenger flow prediction of subway will help to improve people's living standard and save transportation time cost, which is of practical significance to urban rail operating companies.

Modelling and evaluation of the post-hardness and forming limit diagram in the single point incremental hole flanging (SPIHF) process using ANN, FEM and experimental

In the single point incremental hole flanging (SPIHF) process, a sheet material with pre-cut holes is deformed using the SPIF technique to generate a flange, making it an effective approach for low volume manufacturing and quick prototyping. In the case of the SPIHF technique, the post-forming hardness property, the forming limit diagram (FLD), and spring-back phenomena are not completely evaluated. To this end, this paper employs experimental investigation and numerical validation to analyse the impact of SPIHF process parameters like tool diameter, feed rate, spindle speed, and initial hole diameter on these aspects for the truncated incrementally formed components made from AA1060 aluminium alloy and DC01 carbon steel. The plasticity behaviour of both sheet metals was simulated using the Workbench LS-DYNA model and ANSYS software version 18. Additionally, Cowper Symonds power-law hardening was added to the model to account for material properties. The average post-hardness of AA1060 and DC01 was evaluated using an SPIHF prediction model based on the performance of an artificial neural network (ANN). This ANN model was developed using a feed-forward back-propagation network trained using the Levenberg-Marquardt approach. The ANNs 4-n-1 were created by varying the transfer functions and the number of hidden neurons. Greater spindle speed and bigger pre-cut holes were shown to significantly increase the post-formed hardness of the truncated components, whereas the converse was seen when using a higher feed rate and a larger tool diameter. In addition, the FLD and spring-back improved dramatically with larger hole diameters. Employing correlation coefficient (R) and mean square error (MSE) as validation measures, it was shown that the established ANN models accurately predicted the SPIHF process response. Both the DC01 and AA1060 neural network models with a 4-8-1 network architecture performed very well, with MSE and R values of 0.0000105 and 1 for DC01 and 0.02613 and 0.99982 for AA1061.

Neural network-based method for determining vessel position by seabed relief

A seabed relief-based vessel position fixing model on the basis of a four-layered feedforward neural network is proposed. Hidden neurons have hyperbolic tangent activation functions. The model is constructed for 1-D case that can be considered as vessel motion throw a narrow channel or alongside fairway axis. A sequence of spot soundings is given for the network input. The linear coordinate registered for the last sounding forms the network output. The training set is formed by means of the intentional pseudorandom alteration of input samples in accordance with suspected limits of sea level variations and the constant error of its measurements. The validation set is not used. The Adamax algorithm is implemented for the neural network training. The maximum of absolute value of the prediction error is used as a performance criterion of the net. Modeling has been conducted with the Python programming language. The Tensorflow library is used for the creation, training and testing of the neural network. The depth is modelled as a piecewise polynomial function of the coordinate. The results of neural network testing with the use of noised input samples let to state that the neural net can determine a ship position by means of soundings with acceptable accuracy. Different combinations of the sea level error and the number of hidden neurons have been considered. For each of such combinations the network accuracy indicators have been calculated. The best results are obtained for the network with 100 hidden neurons per each layer.

Wholesale Food Price Index Forecasts with the Neural Network

Food price forecasts in the agricultural sector have always been a vital matter to a wide variety of market participants. In this work, we approach this forecast problem for the weekly wholesale food price index in the Chinese market during a 10-year period of January 1, 2010–January 3, 2020. To facilitate the analysis, we propose the use of the nonlinear auto-regressive neural network. Technically, we investigate forecast performance, based upon the relative root mean square error (RRMSE) as the evaluation metrics, corresponding to one hundred and twenty settings that cover different algorithms for model estimations, numbers of hidden neurons and delays, and ratios for splitting the data. Our experimental result suggests the construction of the neural network with three delays and 10 hidden neurons, which is trained through the Levenberg–Marquardt algorithm, as the forecast model. It leads to high accuracy and stabilities with the RRMSEs of 1.93% for the training phase, 2.16% for the validation phase, and 1.95% for the testing phase. Comparisons of forecast accuracy between the proposed model and some other machine learning models, as well as traditional time-series econometric models, suggest that our proposed model leads to statistically significant better performance. Our results could benefit different forecast users, such as policymakers and various market participants, in policy analysis and market assessments.

DETECTION OF U2R ATTACKS BY MEANS OF A MULTILAYER NEURAL NETWORK

As a research method, multi layer neural network (MLNN) configurations 41-1-Х-4 were used, where 41 is the number of input neurons; 1 – the number of hidden layers; X – the number of hidden neurons; 4 – the number of resultant neurons created using the Ne

Spatial Heterogeneity Models of Undernourished Toddlers in Java Island, Indonesia

The percentage of undernourished toddlers on Java Island averages 8.5 %. This number is relatively high compared to the average incidence of undernutrition in toddlers in Sumatra, 3.3%. This study uses OLS, GWR, and GWANN to explore undernutrition in toddlers data on Java Island. We use cross-validation to determine the optimum bandwidth in GWR and GWANN methods. GWANN uses a single hidden layer with several hidden neurons where the number of hidden neurons is selected based on RMSE. We use indicators RMSE, MAE, and MAPE to determine the best method for modeling data on undernourished toddlers in Java. By comparing the final results, the best method for modeling data on undernourished toddlers in Java is GWR. There is a relationship between regions on the influence factor of undernutrition. Most adjacent areas, ranging from East Java to DKI Jakarta, have the same influence factor: the percentage of skinny toddlers.

Scrap steel price forecasting with neural networks for east, north, south, central, northeast, and southwest China and at the national level

ABSTRACT There is little doubt about importance of forecasting commodity prices to policy makers and diverse varieties of market participants. In this present work, we analyse price forecasting problems for scrap steel for east, north, south, central, northeast, and southwest China and at the national level with daily data during 08/23/2013–04/15/2021. We focus on exploration of usefulness of non-linear auto-regressive neural networks and examine forecasting performance based upon a variety of settings over algorithms for model estimations, numbers of hidden neurons and delays, and ratios for splitting the data into training, validation, and testing phases. We arrive at relatively simple models for the seven price series, which lead to forecasts of high accuracy and stabilities with relative root mean square errors below 0.85%. Our results could serve as standalone technical forecasts. They could be combined with other forecasts for forming perspectives of price trends and carrying out policy analysis.