Towards enhancing stacked extreme learning machine with sparse autoencoder by correntropy

The stacked extreme learning machine (S-ELM) is an advanced framework of deep learning. It passes the ‘reduced’ outputs of the previous layer to the current layer, instead of directly propagating the previous outputs to the next layer in traditional deep learning. The S-ELM could address some large and complex data problems with a high accuracy and a relatively low requirement for memory. However, there is still room for improvement of the time complexity as well as robustness while using S-ELM. In this article, we propose an enhanced S-ELM by replacing the original principle component analysis (PCA) technique used in this algorithm with the correntropy-optimized temporal PCA (CTPCA), which is robust for outliers rejection and significantly improves the training speed. Then, the CTPCA-based S-ELM performs better than S-ELM in both accuracy and learning speed, when dealing with dataset disturbed by outliers. Furthermore, after integrating the extreme learning machine (ELM) sparse autoencoder (AE) method into the CTPCA-based S-ELM, the learning accuracy is further improved while spending a little more training time. Meanwhile, the sparser and more compact feature information are available by using the ELM sparse AE with more computational efforts. The simulation results on some benchmark datasets verify the effectiveness of our proposed methods.     

Data filtering based maximum likelihood extended gradient method for multivariable systems with autoregressive moving average noise

For multivariable systems with autoregressive moving average noises, we decompose the multivariable system into m subsystems (m denotes the number of outputs) and present a maximum likelihood generalized extended gradient algorithm and a data filtering based maximum likelihood extended gradient algorithm to estimate the parameter vectors of these subsystems. By combining the maximum likelihood principle and the data filtering technique, the proposed algorithms are effective and have computational advantages over existing estimation algorithms. Finally, a numerical simulation example is given to support the developed methods and to show their effectiveness.     

A hierarchical least squares identification algorithm for Hammerstein nonlinear systems using the key term separation

Mathematical models are basic for designing controller and system identification is the theory and methods for establishing the mathematical models of practical systems. This paper considers the parameter identification for Hammerstein controlled autoregressive systems. Using the key term separation technique to express the system output as a linear combination of the system parameters, the system is decomposed into several subsystems with fewer variables, and then a hierarchical least squares (HLS) algorithm is developed for estimating all parameters involving in the subsystems. The HLS algorithm requires less computation than the recursive least squares algorithm. The computational efficiency comparison and simulation results both confirm the effectiveness of the proposed algorithms.     

Maximum likelihood recursive least squares estimation for multivariate equation-error ARMA systems

This paper focuses on the parameter estimation problems of multivariate equation-error systems. A recursive generalized extended least squares algorithm is presented as a comparison. Based on the maximum likelihood principle and the coupling identification concept, the multivariate equation-error system is decomposed into several regressive identification models, each of which has only a parameter vector, and a coupled subsystem maximum likelihood recursive least squares identification algorithm is developed for estimating the parameter vectors of these submodels. The simulation example shows that the proposed algorithm is effective and has high estimation accuracy.     

Auxiliary model based recursive generalized least squares identification algorithm for multivariate output-error autoregressive systems using the decomposition technique

This paper focuses on the parameter estimation problem of multivariate output-error autoregressive systems. Based on the decomposition technique and the auxiliary model identification idea, we derive a decomposition based auxiliary model recursive generalized least squares algorithm. The key is to divide the system into two fictitious subsystems, the one including a parameter vector and the other including a parameter matrix, and to estimate the two subsystems using the recursive least squares method, respectively. Compared with the auxiliary model based recursive generalized least squares algorithm, the proposed algorithm has less computational burden. Finally, an illustrative example is provided to verify the effectiveness of the proposed algorithms.     

Effectiveness analysis of a mixed rumor-quelling strategy

Circulating the truth and quarantining a subset of rumor spreaders are two major rumor-quelling strategies. In practice, a mixture of the two strategies may be more effective than any one of the two strategies. This paper focuses on effectiveness analysis of the mixed strategy. For this purpose, we are going to establish a rumor-truth competing model on two-layer network. First, we introduce a Markov model characterizing the stochastic dynamics of the rumor-truth competing process, and write the corresponding Kolmogorov model capturing the expected dynamics of the rumor-truth competing process. Second, we give a bilinear model as the first approximation to the Kolmogorov model, and suggest a generic model as a more accurate approximation to the Kolmogorov model. The two models are the focus of concern in this work. For ease in treatment, we describe a limit system of the generic model. By studying the limit model, we present a criterion for the rumor to subside, a criterion for the rumor not necessarily to subside, and a criterion for the rumor to persist, respectively. These findings are instructive to the quelling of false rumors. Finally, through computer experiments we find that when a rumor subsides, the bilinear model is a good approximation to the Kolmogorov model.     

Generalized lag synchronization of multiple weighted complex networks with and without time delay

The generalized lag synchronization of multiple weighted complex dynamical networks with fixed and adaptive couplings is investigated in this paper, respectively. By designing appropriate controller, several synchronization criteria are presented for multiple weighted complex dynamical networks with and without time delay based on the selected Lyapunov functional and inequality techniques. Moreover, an adaptive scheme to update the coupling weights is also developed for ensuring the generalized lag synchronization of multiple weighted complex dynamical networks with and without time delay. Finally, two numerical examples are provided in order to validate effectiveness of the proposed generalized lag synchronization criteria.     

A new Lyapunov functional approach to sampled-data synchronization control for delayed neural networks

This paper discusses the problem of synchronization for delayed neural networks using sampled-data control. We introduce a new Lyapunov functional, called complete sampling-interval-dependent discontinuous Lyapunov functional, which can adequately capture sampling information on both intervals from $r(t?\overline{\tau })$ to $r(t_k?\overline{\tau })$ and from $r(t?\overline{\tau })$ to $r(t_{k+1}?\overline{\tau })$. Based on this Lyapunov functional and an improved integral inequality, less conservative conditions are derived to ensure the stability of the synchronization error system, leading to the fact that the drive neural network is synchronized with the response neural network. The desired sampled-data controller is designed in terms of solutions to linear matrix inequalities. A numerical example is provided to demonstrate that the proposed approaches are effective and superior to some existing ones in the literature.     

Input-to-state stability of impulsive inertial memristive neural networks with time-varying delayed

The property of input-to-state stability (ISS) of inertial memristor-based neural networks with impulsive effects is studied. Firstly, according to the characteristics of memristor and inertial neural networks, the inertial memristor-based neural networks are built. Secondly, based on the impulsive control theory, the average impulsive interval approach, Halanay differential inequality, Lyapunov method and comparison property, some sufficient conditions ensuring ISS of the inertial memristor-based neural networks under impulsive controller are derived. In this paper, we consider two types of impulse, stabilizing impulses and destabilizing impulses. When the inertial memristor-based neural networks are originally not ISS, by choosing a suitable lower bound of the average impulsive interval, the stabilizing impulses can be used to stabilize the inertial memristor-based neural networks. On the contrary, the inertial memristor-based neural networks are originally ISS, by restricting the upper bound of the average impulsive interval, the ISS of inertial memristor-based neural networks with destabilizing impulses can be ensured. Finally, numerical results are presented to illustrate the main results.