Chaotic Dynamics in Weight Space of Neural Networks

When a special nonlinear self-feedback term is introduced into the dynamical equation of the backpropagation training algorithm for networks, the dynamics in weight space of networks can become chaotic. Chaotic dynamics of the system can help it escape from the most commonplace local minima of the energy. Simulation on the XOR problem and the prediction of chaotic time series have shown that the proposed chaotic training algorithm can converge to the global minimum or its approximate solutions efficiently and dramatically faster than the original backpropagation training algorithm.     

Self-Organization of Weighted Networks in Connection with the Misanthrope Process

Prom an undirected random graph, by the weight redistribution of the edges, we obtain a weighted network. The weight redistribution of the edges can be connected to the well-known Misanthrope process, in which distinguishable particles hop among different urns. Under specific conditions, the condensation phenomena can be observed, i.e., nearly all the edges connect to one vertex in the network. When there is no condensation, by adjusting the parameters, the strength distribution can be scale-free or exponentially decreasing. The numerical results fit well with the analytical ones.     

Estimating Topology of Discrete Dynamical Networks

In this paper, by applying Lasalle＇s invariance principle and some results about the trace of a matrix, we propose a method for estimating the topological structure of a discrete dynamical network based on the dynamical evolution of the network. The network concerned can be directed or undirected, weighted or unweighted, and the local dynamics of each node can be nonidentical. The connections among the nodes can be all unknown or partially known, Finally, two examples, including a Henon map and a central network, are illustrated to verify the theoretical results.     

Estimating Topology of Discrete Dynamical Networks

In this paper, by applying Lasalle's invariance principle and some results about the trace of a matrix, we propose a method for estimating the topological structure of a discrete dynamical network based on the dynamical evolution of the network. The network concerned can be directed or undirected, weighted or unweighted, and the local dynamics of each node can be nonidentical. The connections among the nodes can be all unknown orpartially known. Finally, two examples, including a Hénon map and a central network, are illustrated to verify the theoretical results.     

无线认知网络中拓扑控制方法研究

作为下一代通信网络,无线认知网络已成为当前的研究热点。由于节点的移动性,无线网络拓扑结构动态变化,拓扑控制一直是无线网络的难点问题。通过借鉴移动自组织网络(MANETS)中的拓扑控制方法,提出了无线认知网络中基于博弈论和认知功能相结合的拓扑控制方法。无线认知节点能够通过主动决策调节自身节点位置,在保证网络连通性的基础上实现网络覆盖面积最大。仿真实验结果验证了方法的有效性和收敛性。     

Local Events and Dynamics on Weighted Complex Networks

We examine the weighted networks grown and evolved by local events, such as the addition of new vertices and links and we show that depending on frequency of the events, a generalized power-law distribution of strength can emerge. Continuum theory is used to predict the scaling function as well as the exponents, which is in good agreement with the numerical simulation results. Depending on event frequency, power-law distributions of degree and weight can also be expected. Probability saturation phenomena for small strength and degree in many real world networks can be reproduced. Particularly, the non-trivial clustering coefficient, assortativity coefficient and degree-strength correlation in our model are all consistent with empirical evidences.     

Sensing Enhancement on Social Networks: The Role of Network Topology

Sensing and processing information from dynamically changing environments is essential for the survival of animal collectives and the functioning of human society. In this context, previous work has shown that communication between networked agents with some preference towards adopting the majority opinion can enhance the quality of error-prone individual sensing from dynamic environments. In this paper, we compare the potential of different types of complex networks for such sensing enhancement. Numerical simulations on complex networks are complemented by a mean-field approach for limited connectivity that captures essential trends in dependencies. Our results show that, whilst bestowing advantages on a small group of agents, degree heterogeneity tends to impede overall sensing enhancement. In contrast, clustering and spatial structure play a more nuanced role depending on overall connectivity. We find that ring graphs exhibit superior enhancement for large connectivity and that random graphs outperform for small connectivity. Further exploring the role of clustering and path lengths in small-world models, we find that sensing enhancement tends to be boosted in the small-world regime.     

Discrete Opinion Dynamics on Online Social Networks

This paper focuses on the dynamics of binary opinions {+1,-1} on online social networks consisting of heterogeneous actors.In our model,actors update their opinions under the interplay of social influence and selfaffirmation,which leads to rich dynamical behaviors on online social networks.We find that the opinion leading to the consensus features an advantage of the initially weighted fraction based on actors' strength over the other,instead of the population.For the role of specific actors,the consensus converges towards the opinion that a small fraction of high-strength actors hold,and individual diversity of self-affirmation slows down the ordering process of consensus.These indicate that high-strength actors play an essential role in opinion formation with strong social influence as well as high persistence.Further investigations show that the initial fraction of high-strength actors to dominate the evolution depends on the heterogeneity of the strength distribution,and less high-strength actors are needed in the case of a smaller exponent of power-law distribution of actors' strength.Our study provides deep insights into the role of social influence and self-affirmation on opinion formation on online social networks.     

Opinion Dynamics on Complex Networks with Communities

The Ising or Potts models of ferromagnetism have been widely used to describe locally interacting social or economic systems. We consider a related model, introduced by Sznajd to describe the evolution of consensus in the scale-free networks with the tunable strength （noted by Q） of community structure. In the Sznajd model, the opinion or state of any spins can only be changed by the influence of neighbouring pairs of similar connection spins. Such pairs can polarize their neighbours. Using asynchronous updating, it is found that the smaller the community strength Q, the larger the slope of the exponential relaxation time distribution. Then the effect of the initial upspin concentration p as a function of the final all up probability E is investigated by taking different initialization strategies, the random node-chosen initialization strategy has no difference under different community strengths, while the strategies of community node-chosen initialization and hub node-chosen initialization are different in fina/probability under different Q, and the latter one is more effective in reaching final state.     

Network Entropy Based on Topology Configuration and Its Computation to Random Networks

A definition of network entropy is presented, and as an example, the relationship between the value of network entropy of ER network model and the connect probability p as well as the total nodes N is discussed. The theoretical result and the simulation result based on the network entropy of the ER network are in agreement well with each other. The result indicated that different from the other network entropy reported before, the network entropy defined here has an obvious difference from different type of random networks or networks having different total nodes. Thus, this network entropy may portray the characters of complex networks better. It is also pointed out that, with the aid of network entropy defined, the concept of equilibrium networks and the concept of non-equilibrium networks may be introduced, and a quantitative measurement to describe the deviation to equilibrium state of a complex network is carried out.     

Discontinuity and Complex Dynamics of Neural Networks

We focus on the discontinuity of a neural network model with diluted and clipped synaptic connections (±l only). The exact evolution rule of the average firing rate becomes a discontinuous piece-wise nonlinear map when very simple functions of dynamical threshold are introduced into the network. Complex dynamics is observed.     

Evolving Networks with Nonlinear Assignment of Weight

We propose a weighted evolving network model in which the underlying topological structure is still driven by the degree according to the preferential attachment rule while the weight assigned to the newly established edges is dependent on the degree in a nonlinear form. By varying the parameter a that controls the function determining the assignment of weight, a wide variety of power-law behaviours of the total weight distributions as well as the diversity of the weight distributions of edges are displayed. Variation of correlation and heterogeneity in the network is illustrated as well.     

Modeling Topology and Nonlinear Dynamical Behavior of the Weighted Scale-Free Networks

An improved weighted scale-free network, which has two evolution mechanisms： topological growth and strength dynamics, has been introduced. The topology structure of the model will be explored in details in this work. The evolution driven mechanism of Olami-Feder Christensen （OFC） model is added to our model to study the self-organlzed criticality and the dynamical behavior. We also.consider attack mechanism and the study of the model with attack is also investigated in this paper. We tlnd there are differences between the model with attack and without attack.     

Modeling Topology and Nonlinear Dynamical Behavior of the Weighted Scale-Free Networks

An improved weighted scale-free network, which has two evolution mechanisms: topological growth and strength dynamics, has been introduced. The topology structure of the model will be explored in details in this work. The evolution driven mechanism of Olami-Feder-Christensen (OFC) model is added to our model to study the self-organized criticality and the dynamical behavior. We also consider attack mechanism and the study of the model with attack is also investigated in this paper. We find there are differences between the model with attack and without attack.     

Relaxation Dynamics in Condensation on Weighted Scale-Free Networks

Most of the realistic networks are weighted scale-free networks. How this structure influences the condensation on it is a challenging problem. Recently, we make a first step to discuss its condensation [Phys. Rev. E 74 （2006） 036101] and here we focus on its evolutionary process of phase transition. In order to show how the weighted transport influences the dynamical properties, we study the relaxation dynamics in a zero range process on weighted scale-free networks. We find that there is a hierarchical relaxation dynamics in the evolution and there is a scaling relation between the relaxation time and the jumping exponent. The relaxation dynamics can be illustrated by a mean-field equation. The theoretical predictions are confirmed by our numerical simulations.     

Sandpile Dynamics Driven by Degree on Scale-Free Networks

We introduce a sandpile model driven by degree on scale-free networks, where the perturbation is triggered at nodes with the same degree. We numerically investigate the avalanche behaviour of sandpile driven by different degrees on scale-free networks. It is observed that the avalanche area has the same behaviour with avalanche size. When the sandpile is driven at nodes with the minimal degree, the avalanches of our model behave similarly to those of the original Bak-Tang-Wiesenfeld （BTW） model on scale-free networks. As the degree of driven nodes increases from the minimal value to the maximal value, the avalanche distribution gradually changes from a clean power law, then a mixture of Poissonian and power laws, finally to a Poisson-like distribution. The average avalanche area is found to increase with the degree of driven nodes so that perturbation triggered on higher-degree nodes will result in broader spreading of avalanche propagation.     

Synchronization and Pattern Dynamics of Coupled Chaotic Maps on Complex Networks

The synchronization and pattern dynamics of coupled logistic maps on a certain type of complex network, constructed by adding random shortcuts to a regular ring, is investigated. For parameters where an isolated map is fully chaotic, the defect turbulence, which is dominant in the regular network, can be tamed into ordered periodic patterns or synchronized chaotic states when random shortcuts are added, and the patterns formed on the complex network can be grouped into two or three branches depending on the coupling strength.