In this paper, the resilient control for linear discrete-time multi-agent networks subjected to unconfined cyber-attacks is investigated based on a data-driven method. Firstly, according to the evolution of the original network dynamics, a distributed data-driven estimation algorithm is presented. On this basis, a switching control law is proposed to solve the resilient consensus problem for the discrete-time multi-agent network under unconfined cyber-attacks. Further, some necessary and sufficient conditions for designing the resilient controllers are obtained by solving a nonlinear matrix inequation. Secondly, the proposed data-driven method is extended to study the resilient tracking control and formation control problems. Finally, some numerical simulations are provided to verify the effectiveness of the data-driven resilient control method.

Attaining the proper balance between underfitting and overfitting is one of the central challenges in machine learning. It has been approached mostly by deriving bounds on generalization risks of learning algorithms. Such bounds are, however, rarely controllable. In this study, a novel bias-variance balancing objective function is introduced in order to improve generalization performance. By utilizing distance correlation, this objective function is able to indirectly control a stability-based upper bound on a model's expected true risk. In addition, the Generalization-Aware Collaborative Ensemble Regressor (GLACER) is developed, a model that bags a crowd of structured regression models, while allowing them to collaborate in a fashion that minimizes the proposed objective function. The experimental results on both synthetic and real-world data indicate that such an objective enhances the overall model's predictive performance. When compared against a broad range of both traditional and structured regression models GLACER was ~10-56% and ~49-99% more accurate for the task of predicting housing prices and hospital readmissions, respectively.

Many spreading processes of information and diseases take place over complex networks that are composed of multiple interconnection layers. The relationship between network structure, nodes' activity and spreading dynamics impose a threshold above which an epidemic endures. The network structure of individual layers can take different forms, such as scale-free or random, which significantly impacts the epidemic threshold. Similarly, the nodes' inter-layer transition dynamics largely influences the threshold as well. In this study we consider an inter-layer dynamics following: a Markov process, and a memory based activity creating inter-event times with a heavy-tail distribution, which are typically observed in human behavior. It is shown that by introducing a layer of inactivity the epidemic threshold can be closely predicted with our previously derived expression for multiplex networks.

Energy efficiency is gaining importance in wireless communication networks which have nodes with limited energy supply and signal processing capabilities. We present a numerical study of cooperative communication scenarios based on simple local rules. This is in contrast to most of the approaches in the literature which enforce cooperation by using complex algorithms and require strategic complexity of the network nodes. The approach is motivated by recent results in evolutionary biology which suggest that, if certain mechanism is at work, cooperation can be favored by natural selection, i. e. even selfish actions of the individual nodes can lead to emergence of cooperative behavior in the network. The results of the simulations in the context of wireless communication networks verify these observations and indicate that uncomplicated local rules, followed by simple fitness evaluation, can generate network behavior which yields global energy efficiency.

We propose a compression method for True Random Bit Generators (TRBGs) that exploits pseudo-chaotic systems. The compression scheme requires extremely low-complex hardware circuits for being implemented whereas its theoretical explanation is based on a weaker and more general interpretation of the Shadowing Theory, focusing on probability measures, rather than on single chaotic trajectories. We prove theoretically how to design the overall compression scheme, in order to assure the final entropy of the compressed TRBG to be arbitrarily close to the maximum theoretical limit of 1 bit/time-step.

In this paper we present a simple, fast, novel algorithm for building networks whose topology has high synchronizability, is robust against failures, and supports efficient communication. We show that the algorithm is able to build these networks in a small number of steps that scales with the networks density. In addition, we track the evolution of different topological properties in the process of generating these networks. The results show that the topological properties are homogeneously distributed and the topology is less authoritative. Furthermore, we show that the networks we generate are more robust than random, geometric random, small-world or scale-free networks with similar average connectivity. Finally, all of the results indicate that the topology of these networks is entangled, which in many cases represents an optimal topology.

In this paper, a novel model for opinion dissemination over network is suggested so that market penetration activities can be described. The model assumes the existence of two different products, one of which is always considered first and the supporters' (or users') opinions are allowed to propagate among individuals of the population. This is to describe the dynamics of market penetration of a particular product, by gaining the market share of its competitor or attracting new users, even the competitor is with higher reputation. From the simulation results, it is found that the product with lower preference can still obtain a non-zero fraction of market share in many cases, when a small-world network is considered. Some level of clustering in the network facilitates the adoption of the product, while increasing randomness undermines its existence. The simulations also show the importance of the users' remembrance of a product and the communication effectiveness in related to market penetration.

The reconstruction of the links among coupled dynamical systems can be handled in many ways. One of them is through synchronization, building a system composed by a "mirror" network and by evolution equations for the links of interest, in such a way that the two networks will synchronize if and only if the links are correctly estimated. A procedure based on this idea to estimate the topology of a network of interacting objects when the state variables are known is already present in literature, and here it is extended to systems ruled by implicit differential equations, to better suit some actual cases arising, in particular, in electric and electronic systems. The developed method is applied to a case of practical interest: the test of connections in a nanoscale device, introduced in the literature as "quantum dot cellular neural network", which can be easily produced in a self-assembled way. The whole iter has some steps sensitive to errors which can lead to a final device where the links among the "dots" are different from the desired ones. Our interest is in applying the method to detect such bad links.

Error-correcting codes based on quasigroup transformations are proposed. For the proposed codes, similar to recursive convolutional codes, the correlation exists between any two bits of a codeword, which can have infinite length, theoretically. However, in contrast to convolutional codes, the proposed codes are nonlinear and almost random: for codewords with large enough length, the distribution of the letters, pair of letters, triple of letters, and so on, is uniform. Simulation results of bit-error probability for several codes in binary symmetric channels are presented.

Synchronization in complex hybrid networks is studied. Hybrid networks presents the so called "small world" phenomenon, i.e. small distance between any pair of nodes and clustering effect, and can be considered as the union of a local and a global graph, the former providing local connections and the latter providing small distances. After recalling a few results stated in previous papers, we prove that although local graph networks do not synchronize when the number N of nodes is large, the addition of only a small number of global edges makes these hybrid networks synchronize. The obtained results are supported by numerical examples.

Synchronization plays central role in processing of information in many systems. In this work we propose a method for both to predict the behavior of the synchronous state of a network of Hindmarsh-Rose neurons and to synthesize such a network with a prescribed time-domain evolution. This method combines the synchronization conditions of the network and the bifurcation diagram of the synchronous state.

Active research in bio-inspired neurons has been observed in the last decade. In this paper, we are interested in identifying the topology of a biological neural network based on an adaptive observer design. It is proved that the topological structure of a network and the connectivities of its neurons can be acquired by observing the neurons' dynamics. That can also be used to monitor and report any changes of the topological structure. The effectiveness of this approach is successfully demonstrated with a network of coupled Hindmarsh-Rose neurons.

Synchronization in networks with different topologies is studied. We show that for a large class of oscillators there exist two classes of networks; class-A: networks for which the condition of stable synchronous state is sigmagamma2> a, and class-B: networks for which this condition reads gammaN/gamma21= 02les... les gammaNare the eigenvalues of the Laplacian matrix, where N is the order of the graph. Synchronization in networks whose topology is described by classical random graphs and power-law random graphs when N rarr infin is investigated in detail

Iterative decoding algorithms may be viewed as high-dimensional nonlinear dynamical systems, depending on a large number of parameters. In this work, we introduce a simplified description of several iterative decoding algorithms in terms of the a posteriori average entropy, and study them as a function of a single parameter that closely approximates the signal-to-noise ratio (SNR). Using this approach, we show that virtually all the iterative decoding schemes in use today exhibit similar qualitative dynamics. In particular, a whole range of phenomena known to occur in nonlinear systems, such as existence of multiple fixed points, oscillatory behavior, bifurcations, chaos, and transient chaos are found in iterative decoding algorithms. As an application, we develop an adaptive technique to control transient chaos in the turbo-decoding algorithm, leading to a substantial improvement in performance. We also propose a new stopping criterion for turbo codes that achieves the same performance with considerably fewer iterations.

We treat the turbo decoding algorithm as a dynamical system parameterized by a single parameter that closely approximates the signal-to-noise ratio (SNR). A whole range of phenomena known to occur in nonlinear systems, like the existence of multiple fixed points, oscillatory behavior, bifurcations, chaos and transient chaos are found in the turbo decoding algorithm. As an application, we have developed a novel stopping criterion based on the average entropy of an information block and propose an adaptive strategy as a function of the SNR.

In this paper we describe a procedure for synthesizing high - dimensional synchronized system. When applied to the problem of communication, the transmitted signal is in this case high-dimensional as from a hyperchaotic process and may be useful for encoding messages and private communications.