Search

The purpose of this study is to investigate whether there are differences in handwritten Chinese signatures on different media including paper and electronic devices. Participants were asked to sign specified names on various types of media and the signatures were scanned or saved digitally for subsequent analysis. In this study, using convolutional neural networks and Siamese neural networks as classifiers and comparators, the performance plunge is revealed and thus considerable dissimilarity between the signatures on different media is implied. To further explore this, cubic Bézier curves are fitted to the signatures using the least square method for quantitative statistical analysis. By analyzing the visual changes in the morphology of strokes, several features of signatures are selected and computed, and the paired t‐test and the Wilcoxon signed‐rank test are implemented, which provides a deeper substantiation and explanation of the findings. [ABSTRACT FROM AUTHOR]

Artificial intelligence is more present than ever, both in our society in general and in science. At the center of this development has been the concept of deep learning, the use of artificial neural networks that are many layers deep and can often reproduce human‐like behavior much better than other machine‐learning techniques. The articles in this collection are some recent examples of its application for X‐ray photon science and crystallography that have been published in Journal of Applied Crystallography. [ABSTRACT FROM AUTHOR]

In this paper, we survey the most recent advances in supervised machine learning (ML) and high‐dimensional models for time‐series forecasting. We consider both linear and nonlinear alternatives. Among the linear methods, we pay special attention to penalized regressions and ensemble of models. The nonlinear methods considered in the paper include shallow and deep neural networks, in their feedforward and recurrent versions, and tree‐based methods, such as random forests and boosted trees. We also consider ensemble and hybrid models by combining ingredients from different alternatives. Tests for superior predictive ability are briefly reviewed. Finally, we discuss application of ML in economics and finance and provide an illustration with high‐frequency financial data. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach for performing beam element analysis that considers the nonlinear deformation behavior of various RHS and SHS sections made of mild to high‐strength steel: the Deep Neural Network Direct Stiffness method (DNN‐DSM), which uses Deep Neural Networks (DNN), a subset of machine learning algorithms and more general artificial intelligence approaches, to predict the nonlinear stiffness matrix terms in a beam element formulation for implementation in the direct stiffness matrix (DSM). These predictions are made using trained DNN models drawn from an extensive pool of geometric‐material nonlinear simulations with additional imperfections (GMNIA) using shell‐based models. Initial implementations of this method are able to predict the nonlinear load‐displacement and moment‐rotation behavior of various cross sections with high accuracy. This combines the precision of shell analysis with the computational efficiency of beam element analysis. Previous publications have already demonstrated the suitability and advantages of this method, albeit on a small scale for local buckling prediction. This work goes beyond previous studies and focuses on finite beam element design. This includes a review of the modeling approaches of DNN‐DSM using experimental and numerical results from literature. [ABSTRACT FROM AUTHOR]

This paper develops a data‐driven autonomous method for detection of fatigue damage and classification of the associated damage risk in mechanical structures, based on ultrasonic signal energy. The underlying concept is built upon attenuation of the signal and stability of the attenuation process. The attenuation provides pertinent information for damage quantification, whereas the stability represents resistance toward the fatigue damage growth. The proposed neural network (NN) model has been trained using the scaled conjugate‐gradient back‐propagation method. The NN model is capable of damage detection and damage classification into five classes of increasing risk. The wavelet transform tool has been used to reduce the noisy pattern of the ultrasonic signal energy by using the associated approximation coefficients. The results show that the proposed method of approximation signal energy can detect and classify the damage with an accuracy of up to ∼98.5%. [ABSTRACT FROM AUTHOR]

In big data and machine learning, energy analytics has shown rapid development in the past decade. With this development in energy analytics, the energy data has exponentially increased well in different areas and domains. Energy computing techniques have been used in various domains, particularly in energy data to handle the data analysis process. This paper focuses on explaining the concept and evolution of various research questions formulation, data extraction, quality valuation, search strategy, study selection and reporting the results in energy‐oriented domains to evaluate energy‐computing techniques. The thermal power plant, power system and smart grid datasets were highly used out of plenty of other datasets discussed and the ANN algorithm is most commonly used to classify the dataset. Different sets of algorithms are used for data analysis, classification, converting structured data to unstructured data, regression, visualization, and forecasting. Journal publications from the year 2012 to 2020 have reviewed and adopted a systematic approach to identify the findings of all relevant research methodologies in the defined research domain. [ABSTRACT FROM AUTHOR]

Functional data analysis is an evolving field focused on analyzing data that reveals insights into curves, surfaces, or entities within a continuous domain. This type of data is typically distinguished by the inherent dependence and smoothness observed within each data curve. Traditional functional data analysis approaches have predominantly relied on linear models, which, while foundational, often fall short in capturing the intricate, nonlinear relationships within the data. This paper seeks to bridge this gap by reviewing the integration of deep neural networks into functional data analysis. Deep neural networks present a transformative approach to navigating these complexities, excelling particularly in high‐dimensional spaces and demonstrating unparalleled flexibility in managing diverse data constructs. This review aims to advance functional data regression, classification, and representation by integrating deep neural networks with functional data analysis, fostering a harmonious and synergistic union between these two fields. The remarkable ability of deep neural networks to adeptly navigate the intricate functional data highlights a wealth of opportunities for ongoing exploration and research across various interdisciplinary areas. This article is categorized under:Data: Types and Structure > Time Series, Stochastic Processes, and Functional DataStatistical Learning and Exploratory Methods of the Data Sciences > Deep LearningStatistical Learning and Exploratory Methods of the Data Sciences > Neural Networks [ABSTRACT FROM AUTHOR]

Summary: The recent crisis caused by COVID‐19 directly affected consumption habits and the stability sof financial markets. In particular, the football industry has been hit hard by this pandemic and therefore has more volatile stock prices. Given this new scenario, further research is needed to accurately estimate the value of the shares of football clubs. In this paper, we estimate an asset pricing model in football clubs with different compositions of risk nature using non‐linear techniques of artificial neural networks. Usually, asset pricing models have been estimated with linear methods such as ordinary least squares. Our results show a precision higher than 90% for all the estimated models, which far exceeds those shown by linear methods in the previous literature. We find that the residual represents about 40% of the variance of the price‐dividend ratio. Long‐term risks follow in importance, and above all, the habit component and its behaviour in the face of changes. The importance of the residual component exists due to a low correlation between the asset price and consumer behaviour, but to a much lesser extent than that shown in previous studies. The estimation carried out with artificial neural networks, both the Deep Learning methods and especially the Quantum Neural Network, opens up new possibilities to estimate more efficiently the pricing of financial assets in the football industry. [ABSTRACT FROM AUTHOR]

Complex industrial process modelling is critically important within the context of industrial intelligence. In recent years, soft sensor techniques based on neural networks have become increasingly popular for modelling nonlinear industrial processes. This paper proposes an integrated framework of neural network modelling and evaluation for nonlinear dynamic processes. This framework achieves an integrated solution for modelling, prediction, evaluation, and network structure parameter selection. It can be applied to noisy sensors and dense data in the time domain. The framework's proposed evaluation mechanism employs two novel evaluation metrics, the variational auto‐encoder (VAE)‐based Kullback–Leibler (KL) divergence metric and the maximum likelihood estimation‐based J metric, which both evaluate the model by mining the statistical properties of the residuals. The framework models the dynamic process with a model order based‐gated recurrent units (MOb‐GRU) neural network and a modified transformer model. Numerical experiments demonstrate that the evaluation mechanism functions properly in scenarios with multiple signal‐to‐noise ratios and multiple noise statistical properties and that the framework produces accurate modelling results. [ABSTRACT FROM AUTHOR]

Volatility is an important element for various financial instruments owing to its ability to measure the risk and reward value of a given financial asset. Owing to its importance, forecasting volatility has become a critical task in financial forecasting. In this paper, we propose a suite of hybrid models for forecasting volatility of crude oil under different forecasting horizons. Specifically, we combine the parameters of generalized autoregressive conditional heteroscedasticity (GARCH) and Glosten–Jagannathan–Runkle (GJR)‐GARCH with long short‐term memory (LSTM) to create three new forecasting models named GARCH–LSTM, GJR‐LSTM, and GARCH‐GJRGARCH LSTM in order to forecast crude oil volatility of West Texas Intermediate on different forecasting horizons and compare their performance with the classical volatility forecasting models. Specifically, we compare the performances against existing methodologies of forecasting volatility such as GARCH and found that the proposed hybrid models improve upon the forecasting accuracy of Crude Oil: West Texas Intermediate under various forecasting horizons and perform better than GARCH and GJR‐GARCH, with GG–LSTM performing the best of the three proposed models at 7‐, 14‐, and 21‐day‐ahead forecasts in terms of heteroscedasticity‐adjusted mean square error and heteroscedasticity‐adjusted mean absolute error, with significance testing conducted through the model confidence set showing that GG–LSTM is a strong contender for forecasting crude oil volatility under different forecasting regimes and rolling‐window schemes. The contribution of the paper is that it enhances the forecasting ability of crude oil futures volatility, which is essential for trading, hedging, and purposes of arbitrage, and that the proposed model dwells upon existing literature and enhances the forecasting accuracy of crude oil volatility by fusing a neural network model with multiple econometric models. [ABSTRACT FROM AUTHOR]

Summary: In this paper, we propose a command governor‐based adaptive control architecture for stabilizing uncertain dynamical systems with not only matched but also unmatched uncertainties and achieving the desired command following performance of a user‐defined subset of the accessible states. In our proposed solution, online least‐squares solutions for the matched and unmatched parameters are obtained through integration method and they are employed in the adaptive control framework. Specifically, the matched uncertainty is identified and its effect upon the system behavior is entirely attenuated. Moreover, using the unmatched uncertainty approximation obtained through radial basis function neural networks, the command governor signal is designed to achieve the desired command following performance of the user‐defined subset of the accessible states. With this command governor‐based model reference adaptive control architecture, the tracking error of the selected states can be made arbitrarily small by judiciously tuning the design parameters. In addition to the analysis of the closed‐loop system stability using methods from the Lyapunov theory, our findings are also illustrated through numerical examples. [ABSTRACT FROM AUTHOR]

This paper presents a novel adaptive optimal control algorithm by combining adaptive dynamic programming with nonlinear model predictive control for unknown continuous‐time affine nonlinear systems. The adaptive optimal control design is realized by the model‐critic‐actor architecture. Model neural network, critic neural network and actor neural network are constructed to approximate the system dynamics, the cost function and the optimal control law respectively. The random initialization of neural networks usually influences the control performance, so three neural networks are initialized properly to obtain the suitable initial values so that the control performance is improved. Especially, actor neural network is initialized to approximate the near‐optimal control law which is obtained from nonlinear model predictive control. The convergence of the proposed algorithm is proved by the Lyapunov theory. Finally, simulation results are provided to illustrate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

The sudden rise in the ability of machine learning methodology, such as deep neural networks, to identify and predict with great accuracy instances of malignant cell growth from radiological images has led prominent developers of this technology, such as Geoffrey Hinton, to hold the view that "[...] we should stop training radiologists." Similar views exist in other contexts regarding the replacement of humans with artificial intelligence (AI) technologies. The assumption in these kinds of views is that deep neural networks are better than human radiologists in that they are more accurate, less costly, and have more predictive power than their human counterparts. In this paper, I argue that these considerations, even if true, are simply inadequate as reasons for us to allocate the kind of trust suggested by Hinton and others to these sorts of artifacts. In particular, I show that if the same considerations were true of something other than an AI device, say a pigeon, we would not have sufficient reason to trust them in the same way as suggested of deep neural networks in a medical setting. If this is the case then these considerations are also insufficient to trust AI enough to replace radiologists. Furthermore, I argue that the reliability of AI methodologies such as deep neural networks—which are at the center of this argument—is something that has not yet been established, and doing so faces fundamental challenges. Because of these challenges, it is not possible to ascribe the level of reliability expected from the deployment of a medical device. So, not only are the reasons cited in favor of the deployment of AI technologies in medical settings not sufficient/adequate even if they are true, but knowing whether they are true or not faces non‐trivial epistemic challenges. If this is so, then we have no good reasons to advocate replacing radiologists with AI methodologies such as deep neural networks. [ABSTRACT FROM AUTHOR]

Summary: In this paper, a novel adaptive PID controller for trajectory‐tracking tasks is proposed. It is implemented in discrete time over a hexacopter, and it takes into consideration the unmanned aerial vehicles (UAVs) nonlinear model. The PID controller is developed following an adaptive neural technique, and its stability is verified by the Lyapunov discrete theory. Besides, the neural identification of the dynamic model of the UAV is presented to backpropagate output errors to adjust PID gains with the purpose of reducing the control errors. The validation of the proposed algorithm is performed through experimental results with a hexacopter. [ABSTRACT FROM AUTHOR]

In this paper, a class of neutral-type neural networks with delays in the leakage term on time scales are considered. By using the Banach fixed point theorem and the theory of calculus on time scales, some sufficient conditions are obtained for the existence and exponential stability of almost periodic solutions for this class of neural networks. The results of this paper are new and complementary to the previously known results. Finally, an example is presented to illustrate the effectiveness of our results. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Modern indoor air quality (IAQ) management plans integrate the use of valid air quality models that accurately predict the dynamics of air quality variations within a considered environment. With rapid advancements in the field of computer sciences that helped improve the capabilities of computational resources and the availability of a wide-ranging spectrum of methodologies that address different aspects of the nonlinearity in a multi-dimensional information domain, there is ample scope for environmental professionals to develop and use hybrid IAQ models. This software review paper presents one such methodology that combines the use of the univariate time series and the radial basis function neural network methods in the development and evaluation of univariate time series based radial basis function neural network hybrid IAQ models for the monitored contaminants of carbon dioxide and carbon monoxide inside a public transportation bus using available software. © 2016 American Institute of Chemical Engineers Environ Prog, 35: 320-324, 2016 [ABSTRACT FROM AUTHOR]

Predicting the moment‐rotation response parameters of semi‐rigid steel connections can be challenging given the multitude of components that contribute to the connection's elastic and plastic deformations. This applies to the popular bolted flush end‐plate beam‐to‐column connections (FEPCs). The literature has highlighted the limitations of current analytical, mechanical, and empirical models in providing accurate predictions. Considering these limitations, the application of machine‐learning methods in structural engineering, such as artificial neural networks (ANN), have gained wide attention recently in addressing problems associated with complex structural deformation and damage phenomena. To that end, the superior nonlinearity of ANNs is employed herein in to predict the response characteristics of FEPCs. A dataset of more than 200 specimens, collected from past experimental programs, is utilized to train the ANN for predicting the elastic stiffness, plastic strength, and posy‐yield stiffness. The paper describes the deduction of response parameters from test data using data fitting, the determination of significant geometric and material features, the ANN architecture and algorithms, and the accuracy metrics of the new model. [ABSTRACT FROM AUTHOR]

Natural languages exhibit many semantic universals, that is, properties of meaning shared across all languages. In this paper, we develop an explanation of one very prominent semantic universal, the monotonicity universal. While the existing work has shown that quantifiers satisfying the monotonicity universal are easier to learn, we provide a more complete explanation by considering the emergence of quantifiers from the perspective of cultural evolution. In particular, we show that quantifiers satisfy the monotonicity universal evolve reliably in an iterated learning paradigm with neural networks as agents. [ABSTRACT FROM AUTHOR]

Abstract: The hearing sense is one of the five senses essential for human life. The hearing sense is required to acquire language, learn how to speak, and gain general knowledge. Most importantly, hearing enables individuals to communicate and socialize with the world around them. Deaf people suffer from hearing loss which affects many aspects of their lives; they cannot communicate or socialize effectively with normal people around them. In addition, the safety of deaf people is compromised in many scenarios because they cannot take the right decision emanating from the process of analyzing sounds in the brain, especially alarming sounds. The treatment of hearing loss is dependent on its level. Medicine fails to treat many cases of severe or profound deafness; therefore, people lose hope to live normally. In this paper, we propose a Deaf Assistance Digital System (DADS) that aims to convert major alarm sounds and words which are not distinguishable by deaf people into recognizable alerts for deaf people. The proposed system strives to help deaf people live normally by alerting them about sounds like baby crying or door bell. Moreover, DADS enhances the safety of deaf people by alerting them about alarming sounds and cautionary words. DADS consists of a speech recognition engine and a sound recognition engine. The speech recognition engine is based on the Soundx algorithm while the sound recognition system is implemented using neural networks. The implemented sound recognition and voice recognition engines have more than 90% accuracy. [ABSTRACT FROM AUTHOR]

The cooperative path following problem of multiple underactuated autonomous underwater vehicles (AUVs) involves two tasks. The first one is to force each AUV to converge to the desired parameterized path. The second one is to satisfy the requirement of a cooperative behavior along the paths. In this paper, both of the tasks have been further studied. For the first one, a simplified path following controller is proposed by incorporating the dynamic surface control (DSC) technique to avoid the calculation of derivatives of virtual control signals. Besides, in order to handle the uncertain dynamics, a new type of neural network (NN) adaptive controller is derived, and then an NN based energy-efficient path following controller is firstly proposed, which consists of an adaptive neural controller dominating in the neural active region and an extra robust controller working outside the neural active region. For the second one, in order to reduce the amount of communications between multiple AUVs, a distributed estimator for the reference common speed is firstly proposed as determined by the communications topology adopted, which means the global knowledge of the reference speed is relaxed for the problem of cooperative path following. The overall algorithm ensures that all the signals in the closed-loop system are globally uniformly ultimately bounded (GUUB) and the output of the system converges to a small neighborhood of the reference trajectory by properly choosing the design parameters. Simulation results validate the performance and robustness of the proposed strategy. [ABSTRACT FROM AUTHOR]

In this paper, we suggest basing the development of classification methods on traditional techniques but approximating them in whole or in part with artificial neural networks (ANNs). Compared to a direct ANN approach, the underlying traditional method is often easier to analyse. Classification failures can be better understood and corrected for, while at the same time faster execution can be obtained due to the parallel ANN structure. Furthermore, such a two‐step design philosophy partly eliminates the 'guesswork' associated with the design of ANNs. The expected gain is thus that the benefits from the traditional field, and the ANN field can be obtained by combining the best features from both fields. We illustrate our approach by working through an explicit example, namely, a Nearest Neighbour classifier applied to a subset of the MNIST database of handwritten digits. Two different approaches are discussed for how to translate the traditional method into an ANN. The first approach is based on a constructive implementation which directly reflects the original algorithm. The second approach uses ANNs to approximate the whole, or part of, the original method. An important part of the approach is to show how improvements can be introduced. In line with the presented philosophy, this is done by extending the traditional method in several ways followed by ANN approximation of the modified algorithms. The extensions are based on a windowed version of the nearest neighbour algorithm. We show that the improvements carry over to the ANN implementations. We further investigate the stability of the solutions by modifying the training set. It is shown that the errors do not change significantly. This also holds true for the ANN approximations providing confidence that the two‐step strategy is robust. [ABSTRACT FROM AUTHOR]

Rendering with full lens model can offer images with photorealistic lens effects, but it leads to high computational costs. This paper proposes a novel camera lens model, NeuroLens, to emulate the imaging of real camera lenses through a data-driven approach. The mapping of image formation in a camera lens is formulated as imaging regression functions (IRFs), which map input rays to output rays. IRFs are approximated with neural networks, which compactly represent the imaging properties and support parallel evaluation on a graphics processing unit (GPU). To effectively represent spatially varying imaging properties of a camera lens, the input space spanned by incident rays is subdivided into multiple subspaces and each subspace is fitted with a separate IRF. To further raise the evaluation accuracy, a set of neural networks is trained for each IRF and the output is calculated as the average output of the set. The effectiveness of the NeuroLens is demonstrated by fitting a wide range of real camera lenses. Experimental results show that it provides higher imaging accuracy in comparison to state-of-the-art camera lens models, while maintaining the high efficiency for processing camera rays. [ABSTRACT FROM AUTHOR]

The present paper concerns with the dynamics of a fractional neural network involving three neurons. Firstly, the bifurcation point is identified for which Hopf bifurcations may occur by taking the system parameter as a bifurcation parameter via the stability analysis of fractional systems. It is indicated that the system parameter can significantly affect the dynamical properties of such network. Secondly, the impact of the order on the bifurcation point is carefully examined. It is found that the occurrence of bifurcation is delayed as the order increases as long as the other system parameters are established. Finally, a numerical example is exploited to verify the efficiency of theoretical results. [ABSTRACT FROM AUTHOR]

In this paper, a novel decentralized adaptive neural control scheme is proposed for a class of interconnected large-scale uncertain nonlinear time-delay systems with input saturation. Radial basis function ( RBF) neural networks ( NNs) are used to tackle unknown nonlinear functions. Then, the decentralized adaptive NN tracking controller is constructed by combining Lyapunov- Krasovskii functions and the dynamic surface control ( DSC) technique, along with the minimal-learning-parameters ( MLP) algorithm. The stability analysis subject to the effect of input saturation constraints are conducted with the help of an auxiliary design system based on the Lyapunov- Krasovskii method. The proposed controller guarantees uniform ultimate boundedness ( UUB) of all of the signals in the closed-loop large-scale system, while the tracking errors converge to a small neighborhood around the origin. An advantage of the proposed control scheme lies in the number of adaptive parameters of the whole system being reduced to one and in the solution of the three problems of 'computational explosion,' 'dimension curse,' and 'controller singularity'. Finally, simulation results along with comparisons are presented to demonstrate the advantages, effectiveness, and performance of the proposed scheme. [ABSTRACT FROM AUTHOR]

Summary: Opportunistic Networks (OppNets) is a system of wirelessly connected nodes in a varying network topology. Routing in OppNets is a challenge. To overcome the problem of routing, an intelligent dynamic strategy to select next best node for forwarding a message is required. This paper proposes an intelligent routing mechanism based on Intelligent Water Drop (IWD) Algorithm which is used in tandem with Neural Networks (NNs) as an optimization technique to solve the problem of routing in such networks. The nature–inspired IWD algorithm provides robustness, whereas the neural network base of the algorithm helps it to make intelligent routing decisions. The weights in the Neural Network model are calculated by IWD Algorithm using training data consisting of inputs that are characteristic parameters of nodes, such as buffer space, number of successful deliveries and energy levels along with transitive parameters such as delivery probabilities. The proposed protocol Intelligent Water Drop Neural Network (IWDNN) is compared with other protocols that use similar ideologies such as MLProph, K‐nearest neighbour classification based routing protocol (KNNR), Cognitive Routing Protocol for Opportunistic Network (CRPO), and Inheritance Inspired Context Aware Routing Protocol (IICAR), as well as the standard protocol Prophet. IWDNN is shown to outperform all other protocols with an average message delivery ratio of 60%, which is a significant improvement of over 10% in comparison to other similarly conceived algorithms. It has one of the lowest latency among the protocols studied, in a range of 3000 to 4000 s, and incurs comparably low overhead costs in the range of 15 to 30. The drop ratios are one of the lowest, staying near six and approaching zero as buffer size is increased. Average amount of time a message stayed in the buffer was the lowest, with a mean of 1600 s. [ABSTRACT FROM AUTHOR]

This paper proposes an automatic classification system for impulsive electromagnetic (EM) noises. The classification is based on the frequency spectrum of EM waves emitted from partial discharges in power apparatus. This concept is validated by confirming the reproducibility of waveforms of EM waves. In order to clarify the effectiveness of our proposal, six defects are considered: a needle-plane electrode, a strain insulator, a pin-type insulator, FDF (Forced Draft Fan), CBP (Condensate Booster Pump), and COOL-P (Cooling Water Pump). Neural networks are used to classify the frequency spectrum of EM waves. As a main result, these sources are well classified with a recognition rate of 80% or better. This is supported by treating 1000 EM pulses. © 2006 Wiley Periodicals, Inc. Electr Eng Jpn, 156(3): 1– 8, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20378 [ABSTRACT FROM AUTHOR]

In recent years, research on intelligent systems is becoming more and more popular in academe. Intelligent systems research mainly involves the field of artificial intelligence. There are many research groups engaged in the research of intelligent systems in the world, and a variety of achievements have been made. In this paper, the main research institutions in China as well as their main research topics are introduced. [ABSTRACT FROM AUTHOR]

Fuzzy logic principles and neural networks, both being computational intelligence technologies, can be combined to produce synergetic effects through the formation of a unified approach which takes advantage of the benefits and at the same time counterbalances the flaws of the two technologies. In this paper, a fuzzy neural approach, which is characterized by its ability to suggest the appropriate adjustment of product quantity from various suppliers with different quality standards in a supply chain network, is presented. This approach is particularly useful in situations where multiple supply chain partners are involved to achieve the common objective of producing products to the best satisfaction of customer demands at the lowest possible cost. To validate the feasibility of this approach, a test has been conducted based on the proposed fuzzy neural approach with the objective of suggesting the appropriate selection of suppliers and the optimal quantity allocated to them to meet the required quality standards. This paper describes the methodology for the deployment of this proposed hybrid approach to enhance the machine intelligence of a supply chain network with the description of a case study to exemplify its underlying principles. [ABSTRACT FROM AUTHOR]

Recently, numerous attempts have been made by researchers to understand the essence of complex phenomena (complex systems). In this paper, we consider biological systems in nature as being among the most complex systems. The purpose of this paper is to propose a method of realizing symbiotic phenomena such as mutual benefit, competition, and exploitation more generally than the Lotka–Volterra equation by using neural networks. © 2002 Wiley Periodicals, Inc. Electr Eng Jpn, 140(1): 77–88, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.1172 Copyright © 2002 Scripta Technica [ABSTRACT FROM AUTHOR]

This paper proposes a quantizer based on the Simple Hysteresis Network (SHN). The SHN is a hysteresis neural network in which all of the mutually coupled loads (neurons having hysteresis characteristics) are simplified. An A-D converter proposed by J. J. Hopfield uses neural networks, but this is highly initial-value-dependent so that the output for an arbitrary input is not necessarily satisfactory. The quantizer proposed in this paper guarantees the best output for an arbitrary analogue input. The proposed quantizer requires N wiring lines to obtain N bits of the output. This is significantly less than the A-D converter proposed by Hopfield, and practical circuits can easily be designed. © 1999 Scripta Technica, Electron Comm Jpn Pt 3, 83(4): 1–8, 2000 [ABSTRACT FROM AUTHOR]

This paper discusses techniques that might be helpful in predicting interest rates and tries to evaluate a new hybrid forecasting approach. Results of examining government bond yields in Germany and France reported in this study indicate that a hybrid forecasting approach which combines techniques of cointegration analysis with neural network (NN) forecasting models can produce superior results to the use of NN forecasting models alone. The findings documented in this paper could be a consequence of the fact that examining differenced data under certain conditions will lead to a loss of information and that the inclusion of the error correction term from the cointegration model can help to cope with this problem. The paper also discusses some possibly interesting directions for further research. [ABSTRACT FROM AUTHOR]

In this paper, the authors present a neural-network-based distributed secondary control to regulate the output voltage and frequency of a smart autonomous microgrid system. Generally, the secondary controller is implemented in a centralized manner using a fixed-gain proportional-plus-integral controller which may perform well under certain operating conditions only. Also the failure of centralized controller implies no secondary control action for the entire system. The control technique proposed in this paper is a distributed one and makes use of neural network (NN) concept to improve the performance of system. A well-trained NN supplies the controller with suitable gains according to each operating point. Before training the NN, evolutionary optimization technique, differential evolution, is employed to obtain the optimal gains of controller at each operating load condition which forms the training set for NN. Simulation results show that the proposed controller damps the oscillations caused by load changes, restores the output voltage and frequency of the system to their nominal values, and maintains proper load sharing property of the baseline controller. The performance of the controller is also compared with fixed-gain controller. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Within the evolving landscape of fifth‐generation (5G) wireless networks, the introduction of network‐slicing protocols has become pivotal, enabling the accommodation of diverse application needs while fortifying defences against potential security breaches. This study endeavours to construct a comprehensive network‐slicing model integrated with an attack detection system within the 5G framework. Leveraging software‐defined networking (SDN) along with deep learning techniques, this approach seeks to fortify security measures while optimizing network performance. This undertaking introduces network slicing predicated on SDN with the OpenFlow protocol and Ryu control technology, complemented by a neural network model for attack detection using deep learning methodologies. Additionally, the proposed convolutional neural networks‐long short‐term memory approach demonstrates superiority over conventional ML algorithms, signifying its potential for real‐time attack detection. Evaluation of the proposed system using a 5G dataset showcases an impressive accuracy of 99%, surpassing previous studies, and affirming the efficacy of the approach. Moreover, network slicing significantly enhances quality of service by segmenting services based on bandwidth. Future research will concentrate on real‐world implementation, encompassing diverse dataset evaluations, and assessing the model's adaptability across varied scenarios. [ABSTRACT FROM AUTHOR]

Predicting corporate failure or bankruptcy is one of the most important problems facing business and government. The recent Savings and Loan crisis is one example, where bankruptcies cost the United States billions of dollars and became a national political issue. This paper provides a 'meta analysis' of the use of neural networks to predict corporate failure. Fifteen papers are reviewed and compared in order to investigate 'what works and what doesn't work'. The studies are compared for their formulations including aspects such as the impact of using different percentages of bankrupt firms, the software they used, the input variables, the nature of the hidden layer used, the number of nodes in the hidden layer, the output variables, training and testing and statistical analysis of results. Then the findings are compared across a number of dimensions, including, similarity of comparative solutions, number of correct classifications, impact of hidden layers, and the impact of the percentage of bankrupt firms. [ABSTRACT FROM AUTHOR]

This paper considers on-line learning of neural networks for an inverse problem. A neural network must learn an inverse system of a static unknown system on the basis of an evaluation function that is a squared error between an output of the unknown system and its desired value. Usual learning methods need information about the sign of the sensitivity of the unknown system. This paper proposes learning rules that do not require the information about that sign. These rules utilize a perturbation signal. Rules with the perturbation on an input of the unknown system and a rule with it on each weight of the neural network are presented. The usefulness of these learning rules is confirmed via some numerical simulations. [ABSTRACT FROM AUTHOR]

In this paper, by starting from basic quaternion algebra properties and algorithms, a quaternion‐valued Cohen‐Grossberg neural network was derived, subsequently, several new sufficient conditions are derived to ensure existence and global asymptotic stability (GAS) and global exponential stability (GES) of the equilibrium point (EP) for quaternion‐valued Cohen‐Grossberg neural networks. The obtained criteria can be checked easily in practice and have a distinguished feature from previous studies. Finally, we have numerical evidences that the mathematical system and the conclusions presented are validated. [ABSTRACT FROM AUTHOR]

Summary: This paper proposes distributed adaptive cooperative control algorithms for second‐order agents to track a leader with unknown dynamics. The models of the followers and the leader are composed of uncertain nonlinear components. The order of the leader's dynamics is unknown and can be fractional. Only the single output information is shared among neighbored agents. To simplify the control design, linearly parameterized neural networks are used to approximate the unknown functions. We first present an adaptive control for leaderless consensus and then extend the method to the tracking problem. Thorough theoretical proofs as well as numerical simulation are included to verify the results. Compared with relevant literature, the new approach applies to a larger variety of systems because (i) knowledge about the structure of leader's model is unnecessary; (ii) the unknown functions in different agents' dynamics can be diverse and arbitrary, in other words, the algorithms apply to heterogeneous agents; (iii) the results can be simply used without parameter calculations. [ABSTRACT FROM AUTHOR]

Summary: This paper introduces a multiple‐input–single‐output (MISO) neuro‐fractional‐order Hammerstein (NFH) model with a Lyapunov‐based identification method, which is robust in the presence of outliers. The proposed model is composed of a multiple‐input–multiple‐output radial basis function neural network in series with a MISO linear fractional‐order system. The state‐space matrices of the NFH are identified in the time domain via the Lyapunov stability theory using input‐output data acquired from the system. In this regard, the need for the system state variables is eliminated by introducing the auxiliary input‐output filtered signals into the identification laws. Moreover, since practical measurement data may contain outliers, which degrade performance of the identification methods (eg, least‐square–based methods), a Gaussian Lyapunov function is proposed, which is rather insensitive to outliers compared with commonly used quadratic Lyapunov function. In addition, stability and convergence analysis of the presented method is provided. Comparative example verifies superior performance of the proposed method as compared with the algorithm based on the quadratic Lyapunov function and a recently developed input‐output regression‐based robust identification algorithm. [ABSTRACT FROM AUTHOR]

The evaporation of water from free‐water surfaces or land surfaces is one of the main components of the hydrological cycle and is a complex outcome of various meteorological and physical parameters. A simulation procedure was evaluated in this paper to study the capabilities of a neuro‐fuzzy (NF) technique to estimate daily pan evaporation (EP) magnitudes using meteorological variables. The assessment of the NF technique for simulating EP values was performed via local (temporal) and external (spatial) data‐management scenarios. Hence, a thoroughgoing scan of the possible train and test‐set combinations was carried out based on the temporal and spatial criteria using k‐fold testing procedures. A comparison was also made between the NF and neural networks (NN) methods using the same data and procedures. The obtained outcomes revealed that the proposed generalized NF and NN models have good abilities at simulating EP values using meteorological data. Therefore, the calibration of local NF models might not be required if sufficient meteorological parameters exist in other weather stations. The results also demonstrated that a proper assessment of the models' performance accuracy should include a complete temporal/spatial scanning of the used patterns. The evaporation of water is an important complex component of the hydrological cycle. A modelling study used to assess the capability of neuro‐fuzzy (NF) models to estimate evaporation values using meteorological inputs in low‐altitude stations is reported. Good applicability of the models using exogenous data was confirmed. [ABSTRACT FROM AUTHOR]

We are concerned with the consensus problem for a class of uncertain nonlinear multi‐agent systems (MASs) connected through an undirected communication topology via event‐triggered approaches in this paper. Two distributed control strategies, the adaptive centralized event‐triggered control one and adaptive distributed event‐triggered control one, are presented utilizing neural networks (NNs) and event‐driven mechanisms, where the advantages of the proposed control laws lie that they remove the requirement for exact priori knowledge about parameters of individual agents by taking advantage of NNs approximators and they save computing and communication resources since control tasks only execute at certain instants with respect to predefined threshold functions. Also, the trigger coefficient can be regulated adaptively with dependence on state errors to ensure not only the control performance but also the efficiency of the network interactions. It is proven that all signals in the closed‐loop system are bounded and the Zeno behavior is excluded. Finally, simulation examples are presented for illustration of the theoretical claims. [ABSTRACT FROM AUTHOR]

Summary: This paper is concerned with the problem of state estimation for a class of neural networks with discrete and distributed interval time‐varying delays. We propose a new approach of nonlinear estimator design for the class of neutral‐type neural networks. By constructing a newly augmented Lyapunov‐Krasovskii functional, we establish sufficient conditions to guarantee the estimation error dynamics to be globally exponentially stable. The obtained results are formulated in terms of linear matrix inequalities (LMIs), which can be easily verified by the MATLAB LMI control toolbox. Then, the desired estimators gain matrix is characterized in terms of the solution to these LMIs. Three numerical examples are given to show the effectiveness of the proposed design method. [ABSTRACT FROM AUTHOR]

This paper studies the robust partially mode‐dependent H∞ filtering for nonhomogeneous Markovian jump neural networks with additive gain perturbations. The discrete time‐varying jump transition probability matrix is considered to be a polytope set. A partially mode‐dependent filter with additive gain perturbations is constructed to increase the robustness of the filter, which is subjects to H∞ performance index. Based on the Lyapunov function approach, sufficient conditions are established such that the filtering error system is robustly stochastically stable. The efficiency of the new technique is illustrated by an illustrative example and a biological network example. [ABSTRACT FROM AUTHOR]

In this paper, we consider a class of delayed quaternion‐valued cellular neural networks (DQVCNNs) with impulsive effects. By using a novel continuation theorem of coincidence degree theory, the existence of anti‐periodic solutions for DQVCNNs is obtained with or without assuming that the activation functions are bounded. Furthermore, by constructing a suitable Lyapunov function, some sufficient conditions are derived to guarantee the global exponential stability of anti‐periodic solutions for DQVCNNs. Our results are new and complementary to the known results even when DQVCNNs degenerate into real‐valued or complex‐valued neural networks. Finally, an example is given to illustrate the effectiveness of the obtained results. [ABSTRACT FROM AUTHOR]

Summary: Two‐channel quadrature mirror filter banks can be efficiently established by the combination of real‐valued infinite impulse response all‐pass filters without incurring aliasing and magnitude distortions. The design problem of filter banks is therefore to seek for the phase error minimization of the all‐pass filter coefficients. In this paper, a neural network‐based Lyapunov energy function is used to relate to the objective function of the designed all‐pass filter coefficients. Based on the architecture of neural networks and suitable selection of Hopfield‐related parameters, the all‐pass filter coefficients are obtained when the networks reach convergence. By further using the suitable combination of the designed all‐pass filters, the 2‐channel quadrature mirror filter banks with perfect reconstruction can be efficiently accomplished. Simulation results indicate that the neural network‐based approach has the advantage of satisfactory performance in magnitude and group delay responses in a parallelism manner. The design of 2‐channel quadrature mirror filter bank can be alternatively formulated as a least‐squares phase error optimization using all‐pass filters. A neural network‐based Lyapunov energy function is exploited to relate the objective function of the all‐pass filters optimization. Based on the architecture of neural network and suitable selection of Hopfield‐related parameters, the optimal all‐pass filter coefficients can be obtained when the networks achieve convergence. [ABSTRACT FROM AUTHOR]

Summary: In this paper, a neural network–based adaptive tracking control problem is investigated for a class of nonstrict‐feedback nonlinear systems in the presence of unknown backlash‐like hysteresis nonlinearity, unmodeled dynamics, and unknown control directions. A state feedback controller is developed for the considered system by applying the adaptive backstepping technique and neural networks. The design difficulties exhibited in this paper due to unmodeled dynamics, unknown control directions, and nonstrict‐feedback form are handled by resorting to a dynamic signal, a Nussbaum function, and the variable separation approach, respectively. It is shown that the designed adaptive controller can guarantee that all the signals remain bounded and that the desired signal can be tracked with a small domain of the origin. A numerical example and an example of a real plant for a one‐link manipulator are provided to show the feasibility of the newly designed controller scheme. [ABSTRACT FROM AUTHOR]

SUMMARY In this paper, multiperson zero-sum differential games for a class of continuous-time uncertain nonlinear systems are solved using a new iterative adaptive dynamic programming (ADP) algorithm. The idea is to use ADP technique to obtain the optimal control pair iteratively that makes the performance index function reach the optimal solution of the zero-sum differential games without the system model. It proves that the iterative performance index functions are convergent to the optimal solution of the game. Stability properties of the system under the iterative control pairs are also presented. Neural networks are used to build the system model, approximate the performance index function, and compute the optimal control policy, respectively, for facilitating the implementation of the iterative ADP method. Finally, two simulation examples are given to demonstrate the performance of the proposed method. Copyright © 2012 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

This paper presents a novel system that enables a fully automatic modeling of both 3D geometry and functionality of a mechanism assembly from a single RGB image. The resulting 3D mechanism model highly resembles the one in the input image with the geometry, mechanical attributes, connectivity, and functionality of all the mechanical parts prescribed in a physically valid way. This challenging task is realized by combining various deep convolutional neural networks to provide high‐quality and automatic part detection, segmentation, camera pose estimation and mechanical attributes retrieval for each individual part component. On the top of this, we use a local/global optimization algorithm to establish geometric interdependencies among all the parts while retaining their desired spatial arrangement. We use an interaction graph to abstract the inter‐part connection in the resulting mechanism system. If an isolated component is identified in the graph, our system enumerates all the possible solutions to restore the graph connectivity, and outputs the one with the smallest residual error. We have extensively tested our system with a wide range of classic mechanism photos, and experimental results show that the proposed system is able to build high‐quality 3D mechanism models without user guidance. [ABSTRACT FROM AUTHOR]

Summary: Implantable cardiac pacemaker is a standard medical device to treat heart rhythm disorders. In this paper, a new adaptive backstepping controller is developed to enhance the performance of dual‐sensor pacemakers for regulating the heart rate based on radial basis function neural networks. The robust design of adaptive backstepping controller using Lyapunov functions allows to guarantee the stability and performance of the rate‐adaptive pacing system for accurately accomplishing the heart rate regulation at different preset or desired values. The developed control system has been successfully validated using 12 cases of the preset heart rates for 4 patients during 3 body activities, namely, at rest, walking, and jogging. The resulting root mean square error and maximum error are less than 0.9 and 1.7%, respectively. Moreover, the comparative results of this study showed that the performance of developed backstepping controller is superior to other pacemaker controllers in the previous studies. Therefore, it is potentially valid to be applied in dual‐sensor cardiac pacemakers for the clinical use. [ABSTRACT FROM AUTHOR]

Abstract: In the last decades, neural network approach has often been used to study various and complex engineering problems, such as optimization or prediction. In this paper, a methodology founded on artificial neural networks (ANNs) was used to calculate the stress intensity factor (SIF) in different points of the front of a semi‐elliptical crack present in a rotating shaft, taking into account the shape and depth of the crack, the angle of rotation, and the location of the point in the front. In the event of rotating machines, such as shafts, it is crucial to know the SIF along the crack front because this parameter, according to the Paris Law, is related to the performance of the crack during its propagation. Previously, it was necessary to achieve the data for the ANN training, for this a quasi‐static numerical model was made, which simulates a rotating cracked shaft with a semi‐elliptical crack. The numerical solutions cover a wide range of crack depths and shapes, and rotation angles. The values of the SIF estimated by the ANNs were contrasted with other solutions available in the literature finding a good agreement between them. The proposed neural network methodology is an alternative that offers a very good option for the SIF estimation, because it is efficient and easy to use, does not require high computational costs, and can be used to analyse the propagation of cracks contained in rotating shafts by means of the Paris Law taking into account the nonlinear behaviour of the shaft. [ABSTRACT FROM AUTHOR]

In Energy-Based Neural Networks (EBNNs), relationships between variables are captured by means of a scalar function conventionally called 'energy'. In this article, we introduce a procedure of 'harmony search', which looks for compounds providing the lowest energies for the EBNNs trained on active compounds. It can be considered as a special kind of similarity search that takes into account regularities in the structures of active compounds. In this paper, we show that harmony search can be used for performing virtual screening. The performance of the harmony search based on two types of EBNNs, the Hopfield Networks (HNs) and the Restricted Boltzmann Machines (RBMs), was compared with the performance of the similarity search based on Tanimoto coefficient with 'data fusion'. The AUC measure for ROC curves and 1 %-enrichment rates for 20 targets were used in the benchmarking. Five different scores were computed: the energy for HNs, the free energy and the reconstruction error for RBMs, the mean and the maximum values of Tanimoto coefficients. The performance of the harmony search was shown to be comparable or even superior (significantly for several targets) to the performance of the similarity search. Important advantages of using the harmony search for virtual screening are very high computational efficiency of prediction, the ability to reveal and take into account regularities in active structures, flexibility and interpretability of models, etc. [ABSTRACT FROM AUTHOR]