Stochastic Neural Networks Research Articles

Almost automorphic solutions in distribution for octonion-valued stochastic recurrent neural networks with time-varying delays

In this paper, we consider a class of octonion-valued stochastic recurrent neural networks with time-varying delays whose state variables, self-feedback coefficients, connection weights and external inputs of the networks are all octonion-valued functions. Based on Banach fixed point theorem and inequality technique, we obtain the existence, uniqueness and global exponential stability of almost automorphic solutions in distribution of this kind of neural networks. Our results are new. Finally, an example is given to illustrate the effectiveness of our results.

Synchronization of stochastic memristive neural networks via event‐triggered impulsive control

AbstractThis paper investigates the synchronization problem for master‐slave stochastic memristive neural networks (SMNNs) with time‐delay via event‐triggered impulsive control (ETIC). Firstly, a novel ETIC method is designed for SMNNs to realize quasi‐synchronization in mean square sense. Different from time‐triggered impulsive control (TTIC) in the existing results on SMNNs, the proposed event‐triggered method determines impulsive instants based on the real‐time system states, which can reduce redundant impulses. Compared with the existing ETIC for deterministic memristive neural networks (MNNs), the proposed event‐triggered mechanism (ETM) can exclude Zeno behavior completely by adding the waiting time, when stochastic disturbances occur in MNNs. Secondly, if the master SMNNs have nonzero control inputs, an adaptive‐impulsive hybrid controller with ETM is designed to realize almost surely exponential synchronization. Under the ETM, the number of transmitted events is reduced, then control resources can be saved. Zeno behavior can be excluded in almost sure sense. Ultimately, simulations are presented to illustrate the validity of the obtained results.

A Doubly Stochastic Simulator with Applications in Arrivals Modeling and Simulation

For several areas in the domain of operations research and management science, such as service, logistics and supply chain, and financial systems, the randomness of arrivals is one primary source of uncertainty. Appropriately modeling, statistically characterizing, and efficiently simulating the arrival processes are critical for policy and performance evaluation in the related systems. Classic Monte Carlo simulators have advantages in capturing the interpretable “physics” of a stochastic object, whereas neural network–based simulators have advantages in capturing less-interpretable complicated dependence within a high-dimensional distribution. In “A Doubly Stochastic Simulator with Applications in Arrivals Modeling and Simulation,” Zheng, Zheng, and Zhu propose a doubly stochastic simulator that integrates a stochastic generative neural network and a classic Monte Carlo Poisson simulator to utilize the advantages of both. They provide statistical guarantees and demonstrate empirical performances of the proposed methods.

A computational stochastic procedure for solving the epidemic breathing transmission system

This work provides numerical simulations of the nonlinear breathing transmission epidemic system using the proposed stochastic scale conjugate gradient neural networks (SCGGNNs) procedure. The mathematical model categorizes the breathing transmission epidemic model into four dynamics based on a nonlinear stiff ordinary differential system: susceptible, exposed, infected, and recovered. Three different cases of the model are taken and numerically presented by applying the stochastic SCGGNNs. An activation function ‘log-sigmoid’ uses twenty neurons in the hidden layers. The precision of SCGGNNs is obtained by comparing the proposed and database solutions. While the negligible absolute error is performed around 10–06 to 10–07, it enhances the accuracy of the scheme. The obtained results of the breathing transmission epidemic system have been provided using the training, verification, and testing procedures to reduce the mean square error. Moreover, the exactness and capability of the stochastic SCGGNNs are approved through error histograms, regression values, correlation tests, and state transitions.

Mean-square finite-time synchronization of stochastic competitive neural networks with infinite time-varying delays and reaction–diffusion terms

This paper focuses on the mean-square finite-time synchronization (MFTS) of stochastic competitive neural networks with infinite time-varying discrete delays and reaction–diffusion terms (IRSCNNs). Different from other articles, this paper presents a new approach, which does not use finite-time stability theorem but uses integral inequality, Gronwall-type inequality and comparison strategy to study MFTS. In addition, two control schemes are designed: a feedback control scheme and a new adaptive control strategy, which can achieve MFTS of IRSCNNs. Finally, the correctness of the results is verified by examples.

On Asymptotic Properties of Stochastic Neutral-Type Inertial Neural Networks with Mixed Delays

This article studies the stability problem of a class of stochastic neutral-type inertial delay neural networks. By introducing appropriate variable transformations, the second-order differential system is transformed into a first-order differential system. Using homeomorphism mapping, standard stochastic analyzing technology, the Lyapunov functional method and the properties of a neutral operator, we establish new sufficient criteria for the unique existence and stochastically globally asymptotic stability of equilibrium points. An example is also provided, to show the validity of the established results. From our results, we find that, under appropriate conditions, random disturbances have no significant impact on the existence, stability, and symmetry of network systems.

Finite-Time Synchronization for Stochastic Fractional-Order Memristive BAM Neural Networks with Multiple Delays

This paper studies the finite-time synchronization problem of fractional-order stochastic memristive bidirectional associative memory neural networks (MBAMNNs) with discontinuous jumps. A novel criterion for finite-time synchronization is obtained by utilizing the properties of quadratic fractional-order Gronwall inequality with time delay and the comparison principle. This criterion provides a new approach to analyze the finite-time synchronization problem of neural networks with stochasticity. Finally, numerical simulations are provided to demonstrate the effectiveness and superiority of the obtained results.

From Optimal Control to Mean Field Optimal Transport via Stochastic Neural Networks

In this paper, we derive a unified perspective for Optimal Transport (OT) and Mean Field Control (MFC) theories to analyse the learning process for Neural Network algorithms in a high-dimensional framework. We consider a Mean Field Neural Network in the context of MFC theory referring to the mean field formulation of OT theory that may allow the development of efficient algorithms in a high-dimensional framework while providing a powerful tool in the context of explainable Artificial Intelligence.

Fixed‐time synchronization in multilayer networks with delay Cohen–Grossberg neural subnets via adaptive quantitative control

AbstractIn this paper, fixed‐time synchronization of nonlinear stochastic coupling multilayer neural networks is studied. The neural subnets in the multilayer networks are delay Cohen–Grossberg neural networks (DCGNNs). To overcome uncertain factors, we designed an adaptive delay‐dependent controller in synchronization. To describe constraints of communication and other related problems in networks, which are due to limitations for bit rates and bandwidths in communication channels, an adaptive fixed‐time control strategy is purposed by introducing quantization signal input. A theoretical framework about fixed‐time synchronization in multilayer delay Cohen–Grossberg neural networks (MDCGNNs) is established. We find that fixed settling time is related to the scale of MDCGNNs, characteristics of the designed controller parameters, and level of quantization. Finally, the effective of the theoretical framework is validated in an example.

Neuromorphic computing with hybrid CNN–Stochastic Reservoir for time series WiFi based human activity recognition

Wi-Fi Channel State Information (CSI) based human activity recognition (HAR) which using channel disturbances caused by signal reflection is a novel way of environment sensing and motion recognition. The collected channels characteristics are heavily influenced by the environment, human activity patterns and subject’s weight and height. These signal variations reflected from body components are mainly affected by static multipath effects comprises random noise and behave differently in individuals, and thus an active field of research. To reach further for achieving automated real-time classification, lower computational cost and easy adaptability to hardware are necessary. In this work, a CSI-based HAR with hybrid framework, Convolutional Neural Network (CNN)-Stochastic Reservoir (SR) (CNN-SR) has been proposed, enabling a subject adaptable and more efficient hardware implementation with minimal computational complexity. A subcarrier correlation matrix is first computed and portrayed in image without preprocessing based on the reflection of the raw CSI signal induced by human activities at regular intervals, allowing visual observation of whole pattern changes. The time-based features are subsequently extracted through CNN and these feature arrays are then feed into SR which based on stochastic spiking neural network (SSNN) in simple cycle reservoir architecture for template matching. SR offers attractive power savings over typical von Neumann systems, by doing stochastic computations. The proposed method has also been demonstrated that is capable for HAR based on partially captured signals. The signal pattern of each segment can be observed in a single sight and then employed for person-to-person template recognition. This enables HAR with minimal computational complexity and solving the inter-person variability concerns. The results demonstrate that the proposed CNN-SR achieves impressive performance in recognizing human activities and surpasses existing models with an average accuracy of 93.49%.

Synchronization of Takagi–Sugeno Fuzzy Time-Delayed Stochastic Bidirectional Associative Memory Neural Networks Driven by Brownian Motion in Pre-Assigned Settling Time

We are devoted, in this paper, to the study of the pre-assigned-time drive-response synchronization problem for a class of Takagi–Sugeno fuzzy logic-based stochastic bidirectional associative memory neural networks, driven by Brownian motion, with continuous-time delay and (finitely and infinitely) distributed time delay. To achieve the drive-response synchronization between the neural network systems, concerned in this paper, and the corresponding response neural network systems (identical to our concerned neural network systems), we bring forward, based on the structural properties, a class of control strategies. By meticulously coining an elaborate Lyapunov–Krasovskii functional, we prove a criterion guaranteeing the desired pre-assigned-time drive-response synchronizability: For any given positive time instant, some of our designed controls make sure that our concerned neural network systems and the corresponding response neural network systems achieve synchronization, with the settling times not exceeding the pre-assigned positive time instant. In addition, we equip our theoretical studies with a numerical example, to illustrate that the synchronization controls designed in this paper are indeed effective. Our concerned neural network systems incorporate several types of time delays simultaneously, in particular, they have a continuous-time delay in their leakage terms, are based on Takagi–Sugeno fuzzy logic, and can be synchronized before any pre-given finite-time instant by the suggested control; therefore, our theoretical results in this paper have wide potential applications in the real world. The conservatism is reduced by introducing parameters in our designed Lyapunov–Krasovskii functional and synchronization control.

Boundary Controlling Synchronization and Passivity Analysis for Multi-Variable Discrete Stochastic Inertial Neural Networks

The current paper considers discrete stochastic inertial neural networks (SINNs) with reaction diffusions. Firstly, we give the difference form of SINNs with reaction diffusions. Secondly, stochastic synchronization and passivity-based control frames of discrete time and space SINNs are newly formulated. Thirdly, by designing a boundary controller and constructing a Lyapunov-Krasovskii functional, we address decision theorems for stochastic synchronization and passivity-based control for the aforementioned discrete SINNs. Finally, to illustrate our main results, a numerical illustration is provided.

Random periodic oscillations and global mean-square exponential stability of discrete-space and discrete-time stochastic competitive neural networks with Dirichlet boundary condition

The current article explores the affects of space-time discrete stochastic competitive neural networks. In line with a discrete-space and discrete-time constant variation formula, boundedness and stability are addressed to the space-time discrete stochastic competitive neural networks. Notably, the best convergence speed can be computed by a non-linear optimization problem. In the end, random periodic sequences with respect to time variable of the discrete-space and discrete-time stochastic competitive neural networks are discussed. The results indicate that spatial diffusion with non-negative density factors has no effect on the global mean square boundedness and stability and random periodicity of the network model. The current article is precursory in consideration of space-time discrete competitive neural networks.

Breast cancer detection using Histopathology Image with Mini-Batch Stochastic Gradient Descent and Convolutional Neural Network

Cancer is a devastating disease that has far-reaching effects on our culture and economy, in addition to the human lives it takes. Regarding budgetary responsibility, investing just in cancer treatment is not an option. Early diagnosis is a crucial part of the remedy that sometimes gets overlooked. Malignancy is often diagnosed and evaluated using Histopathology Images (HI), which are widely accepted as the gold standard in the field. Yet, even for experienced pathologists, analysing such images is challenging, which raises concerns of inter- and intra-observer variability. The analysis also requires a substantial investment of time and energy. One way that such an examination may be sped up is by making use of computer-assisted diagnostics devices. The purpose of this research is to create a comprehensive cancer detection system using images of breast and prostate histopathology stained with haematoxylin and eosin (H&E). Proposed here is work on improving colour normalisation methods, constructing an integrated model for nuclei segmentation and multiple objects overlap resolution, introducing and evaluating multi-level features for extracting relevant histopathological image and interpretable information, and developing classification algorithms for tasks such as cancer diagnosis, tumor identification, and tumor class labelling. Mini-Batch Stochastic Gradient Descent and Convolutional Neural Network which obtained statistical kappa value for breast cancer histopathology images shows a high degree of consistency in the classification task, with a range of 0.610.80 for benign and low grades and a range of 0.811.0 for medium and high rates. The Support Vector Machine (SVM), on the other hand, shows an almost perfect degree of consistency (0.811.0) across the several breast cancer picture classifications (benign, low, medium, and high).

Neighborhood-based inference and restricted Boltzmann machine for small molecule-miRNA associations prediction.

A growing number of experiments have shown that microRNAs (miRNAs) can be used as target of small molecules (SMs) to regulate gene expression for treating diseases. Therefore, identifying SM-related miRNAs is helpful for the treatment of diseases in the domain of medical investigation. This article presents a new computational model, called NIRBMSMMA (neighborhood-based inference (NI) and restricted Boltzmann machine (RBM)), which we developed to identify potential small molecule-miRNA associations (NIRBMSMMA). First, grounded on known SM-miRNAs associations, SM similarity and miRNA similarity, NI was used to predict score of an unknown SM-miRNA pair by reckoning the sum of known associations between neighbors of the SM (miRNA) and the miRNA (SM). Second, utilizing a two-layered generative stochastic artificial neural network, RBM was used to predict SM-miRNA association by learning potential probability distribution from known SM-miRNA associations. At last, an ensemble learning model was conducted to combine NI and RBM for identifying potential SM-miRNA associations. Furthermore, we conducted global leave one out cross validation (LOOCV), miRNA-fixed LOOCV, SM-fixed LOOCV and five-fold cross validation to assess performance of NIRBMSMMA based on three datasets. Results showed that NIRBMSMMA obtained areas under the curve (AUC) of 0.9912, 0.9875, 0.8376 and 0.9898 ±0.0009 under global LOOCV, miRNA-fixed LOOCV, SM-fixed LOOCV and five-fold cross validation based on dataset 1, respectively. For dataset 2, the AUCs are 0.8645, 0.8720, 0.7066 and 0.8547 ±0.0046 in turn. For dataset 3, the AUCs are 0.9884, 0.9802, 0.8239 and 0.9870 ±0.0015 in turn. Also, we conducted case studies to further assess the predictive performance of NIRBMSMMA. These results illustrated the proposed model is a useful tool in predicting potential SM-miRNA associations.

Interface-enhanced catalytic performance of TiO2-supported Cu and Au for dimethyl oxalate hydrogenation: A comparative microkinetic analysis

The electronic structures of Cu-TiO2 and Au-TiO2 and their catalytic behaviors in the hydrogenation of dimethyl oxalate (DMO) have been studied by the stochastic surface walking-global neural network potential method, density functional theory calculation, and microkinetic analysis. Calculated results indicate that metal clusters on the pristine and oxygen-deficient TiO2 are mostly coordinated to O2c ions, and the defective surface may provide additional oxygen vacancies for Cu6 and Au6 cluster anchoring. It turns out that the interfacial Ti5c and Cuδ+/Auδ- ions provide adsorption and hydrogenation sites for the reaction intermediates with unsaturated O and C atoms, respectively. The turnover frequency for DMO consumption is much higher on Cu-TiO2 than on Au-TiO2, Cu(111), and Cu(211) at 443.15 K and 20 bar of total pressure. Given the satisfactory methyl glycolate (MG) selectivity, the Cu-TiO2 catalyst has the potential to act as an excellent catalyst for the selective MG production.

An Event-Triggered Method for Stabilization of Stochastic Quaternion-Valued Memristive Neural Networks

An Event-Triggered Method for Stabilization of Stochastic Quaternion-Valued Memristive Neural Networks

The Predictive Power of Social Media Sentiment: Evidence from Cryptocurrencies and Stock Markets Using NLP and Stochastic ANNs

Cryptocurrencies are nowadays seen as an investment opportunity, since they show some peculiar features, such as high volatility and diversification properties, that are triggering research interest into investigating their differences with traditional assets. In our paper, we address the problem of predictability of cryptocurrency and stock trends by using data from social online communities and platforms to assess their contribution in terms of predictive power. We extend recent developments in the field by exploiting a combination of stochastic neural networks (NNs), an extension of standard NNs, natural language processing (NLP) to extract sentiment from Twitter, and an external evolutionary algorithm for optimal parameter setting to predict the short-term trend direction. Our results point to good and robust accuracy over time and across different market regimes. Furthermore, we propose to exploit recent advances in sentiment analysis to reassess its role in financial forecasting; in this way, we contribute to the empirical literature by showing that predictions based on sentiment analysis are not found to be significantly different from predictions based on historical data. Nonetheless, compared to stock markets, we find that the accuracy of trend predictions with sentiment analysis is on average much higher for cryptocurrencies.

A security model for smart grid SCADA systems using stochastic neural network

AbstractDetection of cyber‐threats in the smart grid Supervisory Control and Data Acquisition (SCADA) is still remains one of the complex and essential processes need to be highly concentrated in present times. Typically, the SCADA is more prone to the security issues due to their environmental problems and vulnerabilities. Therefore, the proposed work intends to design a new detection approach by integrating the optimization and classification models for smart grid SCADA security. In this framework, the min‐max normalization is performed at first for noise removal and attributes arrangement. Here, the correlation estimation mechanism is mainly deployed to reduce the dimensionality of features by choosing the relevant features used for attack prediction. Moreover, the optimal features are selected by using the optimal solution provided by the Holistic Harris Hawks Optimization (H3O). Finally, the Perceptron Stochastic Neural Network (PSNN) is utilized to categorize the normal and attacking data flow in the network with minimal processing time and complexity. By using the combination of proposed H3O‐PSNN technique, the detection accuracy is improved up to 99% for all datasets used in this study, and also other measures such as precision to 99.2%, recall to 99%, f1‐score to 99.2% increased, when compared to the standard techniques.

The pth Moment Exponential Synchronization of Drive-Response Memristor Neural Networks Subject to Stochastic Perturbations

In this paper, the p t h moment exponential synchronization problems of drive-response stochastic memristor neural networks are studied via a state feedback controller. The dynamics of the memristor neural network are nonidentical, consisting of both asymmetrically nondelayed and delayed coupled, state-dependent, and subject to exogenous stochastic perturbations. The pth moment exponential synchronization of these drive-response stochastic memristor neural networks is guaranteed under some testable and computable sufficient conditions utilizing differential inclusion theory and Filippov regularization. Finally, the correctness and effectiveness of our theoretical results are demonstrated through a numerical example.