Function Neural Network Research Articles

A TaOx-Based RRAM with Improved Uniformity and Excellent Analog Characteristics by Local Dopant Engineering

Resistive random-access memory (RRAM) with the ability to store and process information has been considered to be one of the most promising emerging devices to emulate synaptic behavior and accelerate the computation of intelligent algorithms. However, variation and limited resistance levels impede RRAM as a synapse for weight storage in neural network mapping. In this work, we investigate a TaOx-based RRAM with Al ion local doping. Compared with a device without doping, the device with locally doped Al ion exhibits excellent uniformity and analog characteristics. The operating voltage and resistance states show tighter distributions. Over 150 adjustable resistance states can be achieved through tuning compliance current (CC) and reset stop voltage. Moreover, incremental resistance switching is available under optimized identical pulses. The improved uniformity and analog characteristics can be attributed to the collective effects of reduced oxygen vacancy (Vo) formation energy and weak conductive filaments induced by the local Al ion dopants.

Modeling and Fuzzy Feedforward Control of Fuel Cell Air Supply System

As an important part of the fuel cell subsystem, the air supply system of the proton exchange membrane fuel cell (PEMFC) plays an important role in improving the output performance and durability of fuel cells. It is necessary to control the oxygen excess ratio of fuel cell systems in the process of variable load, preventing the oxygen starvation in the loading process and excessive parasitic power consumption caused by oxygen saturation. At this time, the modeling of fuel cell systems and the development of control strategies are critical. The development of a control strategy depends on the construction of the control model. Aiming at the difficulty of air supply system modeling, this paper uses radial basis function (RBF) neural network and state equation method to establish the dynamic model of air supply systems. At the same time, PID, fuzzy logic plus PID (FL + PID), feedforward plus PID (FF + PID), fuzzy feedforward plus fuzzy PID (FF + FLPID) control strategy are proposed to control the oxygen excess ratio of the system. The simulation results show that fuzzy feedforward plus fuzzy PID (FF + FLPID) has the best effect and the oxygen excess ratio can be followed in 1 s.

Integration of Functional Link Neural Networks into a Parameter Estimation Methodology

In the field of robust design, most estimation methods for output responses of input factors are based on the response surface methodology (RSM), which makes several assumptions regarding the input data. However, these assumptions may not consistently hold in real-world industrial problems. Recent studies using artificial neural networks (ANNs) indicate that input–output relationships can be effectively estimated without the assumptions mentioned above. The primary objective of this research is to generate a new, robust design dual-response estimation method based on ANNs. First, a second-order functional-link-NN-based robust design estimation approach has been proposed for the process mean and standard deviation (i.e., the dual-response model). Second, the optimal structure of the proposed network is defined based on the Bayesian information criterion. Finally, the estimated response functions of the proposed functional-link-NN-based estimation method are applied and compared with that obtained using the conventional least squares method (LSM)-based RSM. The numerical example results imply that the proposed functional-link-NN-based dual-response robust design estimation model can provide more effective optimal solutions than the LSM-based RSM, according to the expected quality loss criteria.

Line Segment-Based Clustering Approach With Self-Organizing Maps

Clustering techniques are used widely in computer vision and pattern recognition. The clustering techniques are found to be efficient with the feature vector of the input image. So, the present paper uses an approach for evaluating the feature vector by using Hough transformation. With the Hough transformation, the present paper mapped the points to line segment. The line features are considered as the feature vector and are given to the neural network for performing clustering. The present paper uses self-organizing map (SOM) neural network for performing the clustering process. The proposed method is evaluated with various leaf images, and the evaluated performance measures show the efficiency of the proposed method.

Deep Self-Organizing Map Neural Networks for Plantar Pressure Image Segmentation Employing Marr-Hildreth Features

Using the plantar pressure imaging analysis method to realize the optimization design of shoe last is still relatively preliminary. The analysis and utilization of imaging data still have problems such as single processing, incomplete information acquisition, and poor processing model robustness. A deep self-organizing map neural network based on Marr-Hildreth filter (dSOM-wh) is developed in this research. The structure and learning algorithms were optimized by learning vector quantization (LVQ) and count propagation (CP). As a kind of Marr-Hildreth filter, Laplacian of Gaussian (LoG) was developed for the preprocessing. The proposed method performed high effectiveness in accuracy (AC) (92.88%), sensitive (SE) (0.8941), and f-measurement (F1) (0.8720) by comparing with ANN, CNN, SegNet, ResNet, and pre-trained inception-v neural networks. The classification-based plantar pressure biomedical functional zoning technologies have potential applications in the comfort shoe production industry.

Adaptive neural network fixed‐time stabilization control for high‐order nonlinear systems

This paper investigates a neural network (NN) adaptive fixed‐time stabilization control problem for high‐order nonlinear systems. Radial basis function NNs (RBFNNs) are used to approximate unknown nonlinear function. A nonlinear filter is designed to solve the issue of “explosion of complexity.” Combining adaptive backstepping design and adding a power integrator, an adaptive NN fixed‐time control method is developed. It is proved that all the signals of the closed‐loop system are bounded in fixed time. Finally, simulation example is provided to confirm the effectiveness of the presented control method.

Phase retrieval based on difference map and deep neural networks

Phase retrieval occurs in many research areas. There are some classical phase retrieval methods such as hybrid input-output (HIO) and difference map (DM). However, phase retrieval results are sensitive to noise, and the reconstructed images always include artefacts. In this paper, we use the DM algorithm together with DNN to get better phase retrieval results. We train one deep neural network using amplitude images and phase images, respectively. First, using DM, we get initial reconstructed amplitude and phase results. Then, using DNN improves both amplitude and phase results. Finally, using the DM algorithm again improves the DNN results further. The numerical experimental results show that using DM gives better results than HIO, and using DNN improves phase information better than just using DNN to train for amplitude information alone. Compared with only using DNN improves amplitude methods, our method using DM plus DNN plus DM yields a better reconstruction performance for both amplitude and phase.

On the evaluation of crude oil oxidation during thermogravimetry by generalised regression neural network and gene expression programming: application to thermal enhanced oil recovery

Enhancing oil recovery using in-situ combustion (ISC) is an attractive alternative, especially for heavy crudes. During ISC, part of the hydrocarbon is pyrolysed/oxidised, which generates heat and deposits fuel in the combustion front. In this study, crude reactions during ISC are modelled after their thermogravimetry thermo-oxidative profiles using advanced machine learning systems. The model inputs include the weight per cent of asphaltenes, resins, and °API gravity of the oil as well as the heating rate and the temperature. Four types of artificial neural networks (ANNs); namely multilayer perceptron (MLP), generalised regression neural network (GRNN), cascade-forward neural network (CFNN), and radial basis function (RBF) neural network, were employed to develop models for accurate prediction of the weight per cent of residual crude oil based on 2289 experimental data points. Moreover, three optimisation algorithms; including Bayesian Regularisation (BR), Levenberg–Marquardt (LM), and Scaled Conjugate Gradient (SCG) were applied in the training step of MLP and CFNN to improve the prediction ability. GRNN provided the most accurate prediction with ∼2.3% overall average absolute per cent relative error and coefficient of determination of 0.9983. GRNN model is reliable for crude oils with °API gravity of 5–35 and up to 820°C. Lastly, a mathematical correlation was developed to estimate the residual crude oil from thermogravimetry analysis using gene expression programming (GEP). GEP also predicted the thermo-oxidative profile with high accuracy. On the basis of sensitivity analysis, residue formation during crude oil oxidation was impacted the most by the temperature, oil °API gravity, and asphaltenes content, respectively. The Leverage approach identified 2.9% of the data points as doubtful.

A drawer-type abdominal window with an acrylic/resin coverslip enables long-term intravital fluorescence/photoacoustic imaging of the liver

Abstract The liver has a unique vascular structure and regional immunosuppressive characteristics closely linked to the occurrence and development of diseases. There are no long-term, large-field, and high-quality imaging methods to simultaneously obtain the structure of blood vessels and movement activities of immune cells in abdominal organs in vivo. This research developed a drawer-type abdominal window with an acrylic/resin coverslip named DAWarc and applied it to the intravital fluorescence/photoacoustic imaging of the liver for over 10 days. The liver lobe was inserted into the drawer holder of the DAWarc to physically fix the liver, which decreased the imaging artifacts. The acrylic/resin material used as the coverslip has a high compatibility for fluorescence/photoacoustic imaging. Through intravital fluorescence/photoacoustic imaging, information on the structure of hepatic lobules, spatial distribution of nanopomegranate labeled Kupffer cells (KCs), the movement behavior of invariant natural killer T cells, and morphology of KCs were obtained. We also used a self-organizing map neural network to detect tumor metastases in the photoacoustic images automatically. Conclusively, the DAWarc model provided a powerful tool for intravital fluorescence/photoacoustic imaging of the liver; it helped us to better understand the structure of hepatic lobules and the distribution and function of immune cells during the occurrence and development of liver diseases.

Optimally manage the energy between electric vehicle charging stations and electricity distribution system: A hybrid technique

AbstractIn this manuscript, the energy trading model is proposed for investigating the cooperative benefits between several electric vehicle charging stations (EVCSs) and integrated energy systems (IES) based on hybrid system. The proposed hybrid system is combination of Radial‐Basis Function Neural Network (RBFNN) and artificial transgender longicorn algorithm (ATLA), hence it is called RBFNN‐ATLA technique. Initially, the RBFNN approach manages the energy between IES and EVCSs. After that, the original issue breaks down into the main energy trade and payment negotiation issue. The energy trading issue and payment negotiation problem can be solved using ATLA approach. This proposed structure may not only diminish the IES cost, however also enlarge the EVCS profit. The uncertainties in electricity and renewable energy prices are modeled using a robust optimization technique. Additionally, the integrated demand response is modeled for maximizing operational performance. The distributed algorithm depends on the proposed technique is evolved for solving the issue of energy trading, ensuring the privacy of players. The proposed algorithm may obtain the global optimal solutions devoid of adjusting the parameter compared with existing algorithm. The proposed system is performed by the matrix laboratory (MATLAB)/Simulink and the performance is evaluated with other existing methods. The RBFNN‐ATLA method reduces the cost up to 3.76%, 7.793%, 8.210% and 9.01% for independent and cooperative mode. The experimental outcomes demonstrate that the proposed system can accurately detect that optimal global solution.

Efficiency of the t-distribution stochastic neighbor embedding technique for detailed visualization and modeling interactions between agricultural soil quality indicators

Dimensionality reduction is important for revealing important details that may be useful in decision-making. Although different dimensionality reduction methods have been applied in several soil-based studies, Kohonen self-organizing map neural network (KSOM-NN) has attracted significant attention from researchers because of the quality of data visualization and interpretation. However, there is a dearth of studies that compare KSOM-NN and other robust data reduction techniques such as the t -distribution stochastic neighbor embedding ( t -SNE) method to improve visualization and interpretation of the relationships between soil quality indicators in agricultural soil. This study compares the above-mentioned methods for characterizing soil quality indicators including particle size distribution, soil organic matter (SOM), cation exchange capacity (CEC), soil reaction (pH), electrical conductivity (EC), zinc (Zn), iron (Fe), manganese (Mn), potassium (K) and phosphorus (P) in agricultural dryland. There were strongly positive associations identified between some of the variables studied for example, clay/Fe (r = 0.95), clay/SOM (r = 0.79) and Mn/Zn (r = 0.90) based on the correlation matrix output. According to the KSOM-NN, the best map size was a 4 by 7 with Quantization error (QE) = 0.108, Topographic error (TE) = 0.875 and Kaski-Lagus error (K-LE) = 9.104. This map only yielded 2 main clusters. As for the t -SNE, applying various perplexity values (i.e. 5, 6, 7, 8, 9 and 10) enabled better visualization of the soil quality indicators as observed from the cluster formations than the KSOM-NN. Ultimately, the t -SNE was considered a better and promising method for assessing the interactions of soil quality indicators and has the potential to appropriately visualise as well as improve the interpretability of soil results by identifying essential features and similarities. The results of this study have good implications for agricultural soil management and decision-making. Moreover, the results are preliminary findings that may be used in the future to verify detailed aspects of soil biogeochemistry within the study area. • t -SNE outperformed KSOM-NN in soil properties dimensionality reduction and visualization. • The t -SNE method is rarely used for data reduction and visualization in soil-based research. • More detailed clusters of the soil quality indicators were observed while using the t -SNE compared to KSOM-NN. • The t -SNE is recommended for future applications in soil quality characterization research.

A Quantitative Quality Assessment Framework for Continuous Raw EEG Data

This paper describes the process of capturing an expert’s knowledge in the form of human understandable rules and then inserting them into a self-organizing map neural network. The Dynamic Cell Structure (DCS) is a form of self-organizing map used for many purposes, including classification and prediction. The DCS is a topology-preserving network that starts with no pre-structure, but assumes a structure once trained. This paper explores applying the DCS to analyse a social science data set that contains some of the many factors that may influence the mortality rate of a region.

Control the collective behaviors in a functional neural network

Specific biophysical neurons are presented to detect different stimuli, and these external exciting signals are encoded to trigger appropriate firing modes and action potentials for signal propagation between neurons in the network. A thermosensitive neuron can estimate the effect of temperature changes on the excitability and firing modes in nervous system, a photocurrent-dependent neuron can be sensitive to the changes of external illumination or light, and an auditory neuron can perceive acoustic wave when the vibration energy is absorbed and converted into field energy in the loop of neural circuits. From biophysical viewpoint, some specific electric components such as thermistor, phototube, and piezoelectric ceramics can be incorporated into neural circuits for activating specific functions, and thus the external stimuli such as heat, light and vibration can be detected because these energy injections can be converted to intrinsic field energy in the neural circuits. In this paper, three kinds of different neural circuits are coupled in a close loop, energy pumping and the stability of phase synchronization are investigated by regulating the properties of coupling channels, furthermore, the noise effect is also estimated. When induction coil is used to couple the neural circuits, phase stability is controlled under magnetic field coupling, and the activation of noise can change the stability of phase synchronization. The intrinsic field energy in the light-dependent neuron is increased with the increase of coupling intensity when voltage coupling via resistor and magnetic field coupling via induction coil are switched on. In case of electric field coupling via capacitor, the energy in the light-dependent neural circuit keeps oscillatory with small amplitude. These results indicate that magnetic field can be the most suitable way for realizing synchronous information encoding between different functional neurons than the electric synapse coupling because continuous pumping of ions can induce magnetic field in the cell, and all the neurons are controlled completely by effective pumping in field energy.

Identification of geographical origins of Cordyceps based on data of amino acids with self-organizing map neural network

In this study, data of amino acids of Cordyceps samples from Qinghai and Tibet was analyzed with self-organizing map neural network. A model of XY-Fused network was established with the content of 8 major amino acids and total amino acids for the identification of geographical origins of Cordyceps from Qinghai and Tibet. It had the prediction accuracy of 83.3% for the test set. In addition, data mining indicated that methionine was a special kind of amino acid in Cordyceps which could serve as a marker to identify its geographical origins. On this basis, the content ratio of methionine to total amino acids was proposed to be a quantifiable indicator to distinguish Cordyceps from Qinghai and Tibet.

Basketball Sports Injury Prediction Model Based on the Grey Theory Neural Network.

Sports injuries will have an impact on the consistency and systemicity of the training process, as well as athlete training and performance improvement. Many talented athletes have had their careers cut short due to sports injuries. Preventing sports injuries is the best way for basketball players to reduce sports injuries. Many coaches and athletes on sports teams, on the other hand, are unaware of the importance of sports injury prevention. They only realize that the body's sports functions are abnormal when it suffers from sports injuries. As a result, this paper proposes a gray theory neural network-based athlete injury prediction model. First, from the standpoint of a single model, the improved unequal interval model is used to predict sports injury by optimizing the unequal interval model in gray theory. The findings show that it is a good predictor of sports injuries, but it is a poor predictor of the average number of injuries. Following that, in order to overcome the shortcomings of a single model, a gray neural network combination model was used. A combination model of the unequal time interval model and BP neural network was determined and established. The prediction effect is significantly improved by combining the gray neural network mapping model and the coupling model to predict the two characteristics of sports injuries. Finally, simulation experiments show that the proposed method is effective.

Predicting Parameters of Heat Transfer in a Shell and Tube Heat Exchanger Using Aluminum Oxide Nanofluid with Artificial Neural Network (ANN) and Self-Organizing Map (SOM)

This study is a model of artificial perceptron neural network including three inputs to predict the Nusselt number and energy consumption in the processing of tomato paste in a shell-and-tube heat exchanger with aluminum oxide nanofluid. The Reynolds number in the range of 150–350, temperature in the range of 70–90 K, and nanoparticle concentration in the range of 2–4% were selected as network input variables, while the corresponding Nusselt number and energy consumption were considered as the network target. The network has 3 inputs, 1 hidden layer with 22 neurons and an output layer. The SOM neural network was also used to determine the number of winner neurons. The advanced optimal artificial neural network model shows a reasonable agreement in predicting experimental data with mean square errors of 0.0023357 and 0.00011465 and correlation coefficients of 0.9994 and 0.9993 for the Nusselt number and energy consumption data set. The obtained values of eMAX for the Nusselt number and energy consumption are 0.1114, and 0.02, respectively. Desirable results obtained for the two factors of correlation coefficient and mean square error indicate the successful prediction by artificial neural network with a topology of 3-22-2.

Clustering of basketball players using self-organizing map neural networks.

Clustering players based on their abilities, a new perspective and an important opportunity to meet needs that in the light of traditional talent identification and player science, which is held periodically and there is not enough time for them to appear. Early recognition of these abilities is a factor influencing the success of sports teams. Artificial neural network (ANN) is a new method of modelling and prediction. The aim of this study was to cluster basketball players based on their individual abilities. For this purpose, Self-Organizing Map (SOM) Neural Networks were used. The data set used by 3000 NBA players for 2011 until 2018 is from the Basketball-Reference site. Each player is assigned 30 attributes to reduce them using the PCA method and the features for each player were reduced to 12 samples. In order to implement a SOM of features and functions in MATLAB software 65% of the data were used as the network training phase and the remaining 35% were used to the test phase. 12 players’ features as network input and output 9 clusters resulting from the combination of features. After simulation using SOM, accuracy parameter with the help of this system were obtained above 95%. The result of the study showed that the performance of the SOM in clustering basketball players was higher than the K-Means algorithm. The network implemented in this article has a faster speed in the training process and generalizability than similar cases.

Recognition of Fault State of RV Reducer Based on self-organizing feature map Neural Network

In order to accurately evaluate the working state of RV reducer, a fault identification method based on the fault identification model established by Self-Organizing Feature Map (SOM) Neural Network is proposed. Firstly, the data measured by the RV reducer test platform are analyzed by wavelet to obtain the wavelet coefficient. Then, combined with the efficiency data of RV reducer, the mean square frequency, center of gravity frequency and frequency variance of the two groups of data are calculated after Fourier transform and power spectrum analysis. After optimization, several eigenvalues are obtained. The eigenvalues are input into the competitive neural network and SOM neural network to establish the fault identification model. Finally, the results of the fault identification model established by the competitive neural network and SOM neural network are compared. The prediction results show that the fault identification model established by SOM neural network can effectively determine the working state of RV reducer.

Active Noise Control Algorithm with Saturated Actuator

Active noise control system has received its attention in various technical field such as headphone, motor vehicle, etc. Meanwhile, filtered-x least mean square (FxLMS) algorithm is conventional linear algorithm used in active noise control system. It assumes that acoustic path from the noise source and control source to target area are linear. However, in actual system, the secondary path including a D/A converter, an amplifier, and an actuator may exhibits nonlinear distortion like saturation effects. To cope with this nonlinear effects, functional link artificial neural network (FLANN) has been proposed. FLANN uses nonlinear function expansion filter with FxLMS based control algorithm to control the nonlinear effect. In this paper, noise reduction performance and convergence speed are improved by modifying the conventional FLANN algorithm by decoupling the linear and nonlinear part of noise signal.

DISCOVERING HIDDEN CLUSTER STRUCTURES IN CITIZEN COMPLAINT CALL VIA SOM AND ASSOCIATION RULE TECHNIQUE

Significant revolution in different organizations chief’s point of view toward customer treating and the level of product presentation or services resulted in redefining the structure of these organizations based on this point of view. The municipal services are very important as well. The strategy of “CRM” which was so successful in the private sector and has been applying as “CiRM” in the public sector of developed countries could be very useful for this achievement. The main goal of citizen management is realizing the citizen’s needs and demands, improving communication through connection with citizens and optimizing it to increase the level of their satisfaction. The government agencies do it based on their idea and point of view cause the citizen are valuable assets in the planning of services and reduction of costs. This study proposes a combined data mining method to discover hidden knowledge in call citizen compliant of the municipality of Tehran. A Self-organizing map neural network was used to identifying and classifying citizen needs based on RFM analysis. It also classified citizen needs into three majors. the result of classification and clustering of SOM has created a new feature to profiled call’s customer to identify temporal-spatial patterns of problems by using an association rule with the Apriori algorithm. The results of this idea demonstrate that accordance of citizens call compliant in a different area and discovering hidden knowledge can facilitate the performance of human recourse in improving services to citizens.