Search

In recent years, a lot of excellent multi-view clustering methods have been proposed. Because most of them need to fuse all views at one time, they are infeasible as the number of views increases over time. If the present multi-view clustering methods are employed directly to re-fuse all views at each time, it is too expensive to store all historical views. In this paper, we proposed an efficient incremental multi-view spectral clustering method with sparse and connected graph learning (SCGL). In our method, only one consensus similarity matrix is stored to represent the structural information of all historical views. Once the newly collected view is available, the consensus similarity matrix is reconstructed by learning from its previous version and the current new view. To further improve the incremental multi-view clustering performance, the sparse graph learning and the connected graph learning are integrated into our model, which can not only reduce the noises, but also preserve the correct connections within clusters. Experiments on several multi-view datasets demonstrate that our method is superior to traditional methods in clustering accuracy, and is more suitable to deal with the multi-view clustering with the number of views increasing over time.

In this paper, stability of a class of stochastic neural networks with switching signal is studied. Firstly, by means of the method of limiting average dwell time, we analyze the stability of switched systems which potentially contain unstable subsystems and stable subsystems simultaneously. Moreover, considering two types of switchings: stabilizing switchings and destabilizing switchings, we adopt time-dependent parameters to give a description of the relationship between two successive activated subsystems. Based on the obtained results for switched systems, some stability criteria for switched neural networks with stochastic disturbances are derived. At last, we present a numerical example to demonstrate the effectiveness of our results.

In this paper, the bipartite synchronization issue for a class of coupled reaction–diffusion networks with antagonistic interactions and switching topologies is investigated. First of all, by virtue of Lyapunov functional method and pinning control technique, we obtain some sufficient conditions which can guarantee that networks with signed graph topologies realize bipartite synchronization under any initial conditions and arbitrary switching signals. Secondly, for the general switching signal and periodic switching signal, a pinning controller that can ensure bipartite synchronization of reaction–diffusions networks is designed based on the obtained conditions. Meanwhile, a directed relationship between coupling strength and control gains is presented. Thirdly, numerical simulation is provided to demonstrate the correctness and validity of the derived theoretical results for reaction–diffusion systems. We briefly conclude our findings and future work.

Networked evolutionary game theory is an important tool to study the emergence and maintenance of cooperation in natural, social, and economical systems. In this paper, we investigate the dynamics and convergence of a generalized networked evolutionary game, i.e., delayed hyper-networked evolutionary game (HNEG), which considers the multi-players in fundamental network game and time delay in strategies simultaneously. Based on the tool of semi-tensor product (STP) of matrices, the definition of delayed potential HNEG and representation of potential function are given. Moreover, we conclude the steps to analyze the dynamics and convergence of delayed potential HNEGs. Considering the efficiency in updating process, we define a new strategy updating rule based on the myopic best response adjustment rule (MBRAR), which is called delayed group-based sequential MBRAR. Furthermore, we prove that delayed potential HNEG converges to one of the pure Nash equilibrium trajectories under this rule. Finally, public good game is provided to illustrate the realistic application of our results.

This paper investigates the problem of global μ -synchronization of impulsive pantograph neural networks. In this paper, new concept of ν -asymptotic periodic impulsive interval T a s y ν is proposed for pantograph networks. By employing the Lyapunov method combined with the mathematical analysis approach for impulsive systems, some useful criteria are derived to guarantee the global μ -synchronization of coupled pantograph neural networks when the asymptotic logarithmic periodic impulsive interval T a s y ln ∞ and T a s y ln = ∞ , respectively. Especially when T a s y ln = ∞ , as long as the networks are unstable, impulsive control cannot achieve synchronization regardless of the size of the impulse gain. Numerical simulations are exploited to illustrate our theoretical results.

Recent advances in machine learning have helped solve many challenges in artificial intelligence applications, such as traffic flow forecasting. Traffic flow forecasting models based on machine learning have recently been widely applied because of their great generalisation capability. This study aims to construct a multi-intersection-aware traffic flow prognostication architecture considering recent information of a nearby road, which is a significant indicator of the near-future traffic flow, and considering the selection of appropriate and essential sensors significantly correlated to the future traffic flow. To capture the inner correlation between sequential traffic flow data, a novel learning method involving the relevance vector machine is employed for the traffic flow forecasting. To optimise the kernel parameters of the relevance vector machine, a combination of the chaos theory and a simulated annealing algorithm is adopted. The proposed model is verified with the real-world data of six roads in a Minnesotan city. Then, the forecasting results of the new model are compared with those of some state-of-the-art models. These results indicate that the application of relevance vector regression to short-term traffic flow forecasting combined with a chaos-simulated annealing algorithm to optimise the corresponding parameters is a high-precision and -scalability short-term traffic flow forecasting method. The multi-intersection-aware mechanism helps improve the forecasting accuracy.

In the paper, synchronization of coupled neural networks with delayed impulses is investigated. In order to overcome the difficulty that time delays can be flexible and even larger than impulsive interval, we propose a new method of average impulsive delay (AID). By the methods of average impulsive interval (AII) and AID, some sufficient synchronization criteria for coupled neural networks with delayed impulses are obtained. We prove that the time delay in impulses can play double roles, namely, it may desynchronize a synchronous network or synchronize a nonsynchronized network. Moreover, a unified relationship is established among AII, AID and rate coefficients of the impulsive dynamical network such that the network is globally exponentially synchronized (GES). Further, we discuss the case that time delays in impulses may be unbounded, which has not been considered in existing results. Finally, two examples are presented to demonstrate the validity of the derived results.

As semiconductor feature size continues to decrease and the density of integration continues to increase, highly reliable circuit design is experiencing many challenges, including reliability evaluation, which is one of the most important steps in circuit design. However, faced with the very large scale of integrated circuits at present, traditional simulation-based methods are slightly inadequate in terms of computational complexity and do not apply to the circuits at the concept stage. To solve this problem, this paper presents a new prediction method for circuit reliability based on deep auto encoder networks. Firstly, we analyze and extract the main features associated with circuit reliability. Next, we construct an efficient method for data collection by combining the characteristics of the feature set with the requirements of deep auto encoder networks. Then, we build a deep auto encoder network model oriented to circuit reliability prediction in a supervised learning manner. Simulation results on 74-series circuits and ISCAS85 benchmark circuits show that although the accuracy of the proposed method is slightly lower than that of both the Monte Carlo (MC) method and the fast probabilistic transfer matrix (F-PTM) model, its time-space consumption is approximately constant on different circuits, and it is 102,458,469 times faster than the MC method, and approximately 4,383 times faster than the F-PTM model. Furthermore, the proposed method could be used to predict circuit reliability at the conceptual stage, and it is a very efficient approximation method that could greatly reduce the power consumption of the calculation.

Collaborative Filtering (CF) algorithms have been widely used to provide personalized recommendations in e-commerce websites and social network applications. Among them, Matrix Factorization (MF) is one of the most popular and efficient techniques. However, most MF-based recommender models only rely on the past transaction information of users, so there is inevitably a data sparsity problem. In this article, we propose a novel recommender model based on matrix factorization and semantic similarity measure. Firstly, we propose a new semantic similarity measure based on semantic information in the Linked Open Data (LOD) knowledge base, which is a hybrid measure based on feature and distance metrics. Then, we make an improvement on the traditional MF model to deal with data sparsity. Specifically, the MF process has been extended from both the user and item sides with implicit feedback information and semantic similar items, respectively. Experiments on two real datasets show that our proposed semantic similarity measure and recommender model are superior to the state-of-the-art approaches in recommendation performance.

Sylvester matrix equations are widely used in many engineering fields. When referred to the time-varying Sylvester, the computational time increases sharply because of the heavy burden calculations for not satisfying the computational requirement. Recently, a type of asymptotic neural networks called Zhang neural network (ZNN) has been proposed to achieve astonished convergent speed for Sylvester computing problems as long as the time goes to infinity. How to make the convergent rate to finite time is worth to think about, which induces us to find a finite-time method to accelerate the convergent speed. To tackle this problem, two types of recurrent neural models named terminal computing neural dynamics (TCND) and its accelerated form (ATCND), which are of terminal attraction characteristics are constructed. The proposed neural dynamics are developed not only to accelerate the convergent speed, but also to improve the convergent accuracy of the dynamic error generated by time-varying systems. The upper bound of the convergent time is given via mathematical deduction. In addition, simulation performances of time-varying Sylvester equations are evaluated by TCND, ATCND and ZNN for comparison. Furthermore, a repeatable trajectory motion scheme based on TCND is derived for the solution of trajectory planning with redundant manipulators. Experimental results validate the effectiveness of the novel neural solving dynamics.

In this paper, the globally exponential synchronization problem of dynamical networks with nonlinearly coupling function is considered. Hybrid pinning control strategies are established to force the states of the network to follow some objective trajectory. The impulsive pinning controllers are used to control a fringe of nodes at the impulsive instants, while during the impulsive instants, pinning state-feedback controllers are designed to achieve the control objective. Finally, the validity of the developed techniques is evidenced by an illustrative example.

In this paper, we investigate the static output feedback set stabilization for context-sensitive probabilistic Boolean control networks (CS-PBCNs) via the semi-tensor product of matrices. An algorithm for finding the largest control invariant set with probability one is obtained by the algebraic representations of logical dynamics. Based on the analysis of the set stabilization, necessary and sufficient conditions for S-stabilization are obtained. Static output feedback controllers are designed to achieve S-stabilization for a CS-PBCN. At last, examples to study metastatic melanoma are given to show the effectiveness of our main results.

Deep learning (DL) techniques have been widely used in recommender systems for user modeling and matching function learning based on historical interaction matrix. However, existing DL-based recommendation methods usually perform static user preference modeling by using historical interacted items of the user. In this article, we present a time-aware deep CF framework which contains two stages: dynamic user preference modeling based on attention mechanism and matching score prediction based on DL. In the first stage, short-term user preferences are modeled by the time-aware attention mechanism that fully considered the predicted item, the recent interacted items and their interaction time. The resulting short-term preferences are combined with long-term preferences for dynamic user preference modeling. In the second stage, high-order user-item feature interactions are learned by two types of DL models, Deep Matrix Factorization (DMF) and Multiple-Layer Perception (MLP), and the feature interaction vectors of the two models are fused in the last layer of the model to predict the matching score. Extensive experiments on five datasets indicate that our method is superior to the existing time-aware and DL-based recommendation methods in top-k recommendations significantly and consistently.

In this paper, the equivalence issue between state feedback controller and free sequence controller in Boolean control network (BCN) is investigated. Based on the algebraic representation of Boolean networks, we prove that a Boolean control network can be stabilized to a cycle or a fixed point via free sequence controller if and only if there is a state feedback controller which can derive the Boolean control network to the cycle or the fixed point. Then, two algorithms are provided to find the state feedback controller. After that, we prove that a BCN can be globally controllable by free sequence controller, while the BCN is not necessarily globally controllable by state feedback controller. At last, an example is given to illustrate the results.

Modern data centers coordinate hundreds of thousands of heterogeneous tasks aiming at providing highly reliable cloud computing services. Failure prediction is of vital importance in the analysis of cloud reliability. Recently, a novel kernel learning method called relevance vector machine (RVM) has been widely applied to solve nonlinear predicting problems and has been verified to perform well in many situations. However, it remains a great challenge for existing approaches to acquire the optimal RVM parameters. In this research, an artificial immune system is introduced into a Quantum-inspired Binary Gravitational Search Algorithm (QBGSA) in order to improve the convergence rate of standard QBGSA. In addition, a hybrid model of RVM with improved QBGSA called IQBGSA-RVM is proposed that aims to predict the failure time of cloud services. To evaluate the effectiveness of IQBGSA-RVM in failure prediction, its predicting performance is compared with that of the following algorithms, all of which employs RVM: chaotic genetic algorithms, binary gravitational search algorithms, binary particle swarm optimization, quantum-inspired binary particle swarm optimization and standard QBGSA. The experimental results show that the IQBGSA-RVM model is either comparable to the other models or it outperforms them, to say the least.

The traditional collaborative filtering (CF) method based on static user preference modeling and linear matching function learning severely limits the recommendation performance. To solve the above problem, in this article, we adopt dynamic user preference modeling and nonlinear matching function learning in the CF recommendation. For dynamic user preference modeling, a two-layer neural attention network is used, which fully considers the predicted item, the recent historical interacted items and their interaction time to estimate the contribution weight of each interacted item in user preferences modeling. For nonlinear matching function learning, we add a single hidden layer neural network on top of the traditional matrix factorization (MF) model, which can significantly improve the feature interaction learning capabilities of the model with only a few additional parameters. Extensive experiments show that our method significantly outperforms the state-of-the-art CF methods and the key technologies we proposed in this research have a positive effect on improving the recommendation performance.

In this paper, we study the set stabilization of Boolean networks (BNs) under pinning control strategy. First, the algebraic expression of BN is obtained by using semi-tensor product of matrices. Based on the algebraic expression, we give a method to choose pinning nodes, and achieve set stabilization by controlling these selected nodes. A matrix is further defined to design state feedback controllers. Based on the matrix set, state feedback controllers can be obtained quickly and the computational complexity can be reduced. Finally, an example is given to illustrate the design procedure of pinning controllers.

The verification of wearable embedded software trustability plays a prominent role in the research of wearable embedded technology. Software testing is an alternative way by which to improve the trustability of embedded software systems. The test suites developed are often reused because of the frequent version updates of wearable embedded software; as a result, many test cases in the test suite become redundant. In this paper, a new Boolean subtraction operation is proposed, one that does not alter the coverage of test requirements by the test suites. Redundant requirements and cases are reduced by a column–row transformation operation through the newly defined operation, and the interrelation among the testing requirements and test cases is taken into consideration. The results of experiments on some wearable embedded software test suite reduction indicate that the proposed method results in improved properties compared with H, GE, GRE, and can generate a minimal test suite with which all test requirements can be sufficiently tested.

In this paper, we propose a deep collaborative filtering recommendation model, which consists of an attention-based representation learning component and a multi-input matching function learning component. This model takes interaction matrix based on implicit feedback as data source to construct representations of long-term user preferences and item latent features. In the representation learning, a time-aware attention network is used, which uses the embedding vectors of the predicted item, recent historical interaction items, and the interaction time of recent historical interaction items to estimate the weights of different historical interaction items to short-term user preferences modeling. Then, the dynamic user preference representation can be obtained by combining short-term preferences with long-term preferences. In the matching function learning, a multi-input deep learning model is used. Its input includes not only the dynamic user preference representation and the item latent feature representation, but also the linear interaction between the two representations, so that the model has more powerful feature interactions learning ability. Experimental results on four datasets from different domains show that our method is largely superior to the state-of-the-art collaborative filtering methods, and the novel techniques we propose in this paper are effective in improving recommendation performance.

Software failure time prediction, which is critical in software dependable evaluation, is among one of the most critical issues in software quality assurance research. Conventional software reliability predictions often use single models to illustrate the overall characteristic of the process of software failure. However, those models often fail to produce good prediction performance owing to the constant changes in software failure patterns during different time periods. Fuzzy systems like Takagi–Sugeno–Kang are multimodel approaches that combine simple submodels to represent the overall characteristic of dynamic nonlinear complex systems. This paper presents a novel Fuzzy system based software reliability prediction model, called regularized extreme learning adaptive neuro-fuzzy inference system, and employs the quantum-inspired binary gravitational search algorithm to determine its suitable regularization parameters. The new model can effectively reduce randomness, computational complexity as well as avoid local optimization when searching for optimal parameters. An experimentation using four real-world software failure data sets was carried out to compare the performance of the proposed model with that of certain representative software reliability prediction models. The results showed that the proposed model with the optimized parameters exhibits superior performance.

Recently, deep learning techniques have been widely used in recommendation tasks and have attained record performance. However, the input quality of the deep learning model has a great influence on the recommendation performance. In this work, an efficient and effective input optimization method is proposed. Specifically, we propose an integrated recommendation framework based on two-stage deep learning. In the first stage, with user and item features as the original input, a low-cost marginalized stacked denoising auto-encoder (mSDA) model is used to learn the latent factors of users and items. In the second stage, the resulting latent factors are combined and used as input vector to the DNN model for fast and accurate prediction. Using the latent factor vector as the input to the deep learning-based recommendation model not only captures the high-order feature interaction, but also reduces the burden of the hidden layer, and also avoids the model training falling into local optimum. Extensive experiments with real-world datasets show that the proposed model shows much better performance than the state-of-the-art recommendation methods in terms of prediction accuracy, parameter space and training speed.

In this paper, consensus problem of general nonlinear multi-agent systems with communication failures is studied based on a distributed consensus algorithm. Under the assumption that the communication graphs are strongly connected, we derive some sufficient consensus criterion for both cases fixed and switched topologies using the relative measured output states of each agent. Moreover, the case that there is a leader in the communication graph is also explicitly investigated. Finally, simulation examples are presented to show the effectiveness of the obtained results.

This paper presents an analytical study of outer synchronization of partially coupled dynamical networks via pinning impulsive controller. At first, more realistic drive-response partially coupled networks are established. Then, based on the regrouping method, some efficient and less conservative synchronization criteria are derived and developed in terms of average impulsive interval. Our results show that, by impulsively controlling a crucial fraction of nodes in the response network, the outer synchronization can be achieved. Finally, illustrated examples are given to verify the effectiveness of the proposed strategy.

In this paper, we focus on the topic of synchronization of Boolean control networks (BCNs) with drive-response structure with impulsive disturbances. By transforming BCNs with impulsive disturbances into its algebraic expression, free control sequence for synchronization of BCNs with impulsive disturbances is designed. Besides, some necessary and sufficient conditions are derived for synchronization of BCNs with impulsive disturbances under feedback control. Moreover, an algorithm is given to design feedback controllers. Lastly, one example is used to better illustrate the derived results.