Search

Concept drifts generally refer to the changing of statistical characteristics of non-stationary series over time, which considerably affect the analysis of time series including prediction, anomaly detection and classification, etc. However, since the external noise interference and internal uncertainty of time series, it is still an open problem to detect the occurrence of concept drifts timely and effectively in real applications. In this article, based on Riemannian manifolds and statistical process control, we propose a novel online algorithm for the concept drift detection of time series. Using the online segmentations with multiple sliding windows, phase space reconstruction of time series is implemented, based on which multi-scale features of series data are calculated. By means of information geometry theory, the obtained features are projected into Riemannian manifolds for the evading of noise interference and structural redundancy in the time series. Finally, with statistical process control, the detection of concept drifts is implemented. The experimental results reveal the promising detection performances verified by both artificial data sets and real-life data sets. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach to audio-visual consistency judgment (AVCJ), in which vowel-like regions exhibiting significant changes in lip shape are used as key pronunciations. Unlike conventional methods that typically apply speech information to the whole utterance and corresponding visual features, the proposed technique involves key pronunciation analysis and frontal lip reconstruction for non-frontal lip frames corresponding to these key pronunciations. This is done to eliminate the influence of variations in video viewing angle on consistency judgment and address the limitations of whole-sentence analysis in a process consisting of four steps. First, key pronunciations are detected in speech utterance. Second, view angle classification is performed on lip frames corresponding to the key pronunciations and then frontal reconstruction is performed by a proposed SL-CycleGAN. Third, deep correlation analysis is applied to the covariance of the deep features obtained from both audio and visual features. Fourth, time-delay and correlation differences are combined to estimate the joint score of each key pronunciation. The results of AVCJ are then acquired by calculating a sentence level score from each joint score. Experimental results from the OuluVS2 bimodal multi-view database showed that the proposed method outperformed several state-of-the-art algorithms, including quadratic mutual information (QMI), space-time canonical correspondence analysis, and multiple variations of SyncNet. [ABSTRACT FROM AUTHOR]

With the implementation of online-service information systems, it is important to detect system anomalies. Logs, serving as the system runtime information, are the key resources to understand system statuses. The existing automatic log anomaly detection approaches focused on the temporal features deriving from sequential nature of logs/events, which can be suitable for limited-sized information systems with homogeneous structure. However, with increasing complexity of systems, the time-aware relationships among discrete logs/events become non-trivial features. Different from the previous works, a novel framework for log anomaly detection is proposed from the perspective of link prediction. First, a temporal event graph is constructed to model the time-aware relationships between events, which is derived from discrete event sequences and possesses the evolutionary properties of sequential events. Second, we pre-train a graph convolutional neural network by using graph contrastive learning to capture local topological features. Third, a temporal convolutional network is used to learn the temporal features. Finally, based on generative adversarial network, link prediction is implemented by generating the next moment event graph, which also is the basis for anomaly detection. Experiments are conducted on public datasets to verify the validity of the model, and superior results are achieved compared to the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

The long-term fluctuation of time series is generally composed of a large number of short-term behaviors with various dynamical characteristics, where kinds of fluctuation patterns in different periods change mutually. In this paper, we propose a novel method to construct fuzzy information granules in polar coordinates and achieve the prediction of long-term fluctuation of time series on the basis of the short-term fluctuation patterns. Firstly, time series are divided into segments by means of the sliding time windows, and fuzzy information granules are defined based on the regression models to indicate the fluctuation patterns of segments of time series. The transfers among different information granules form a dynamical network containing rich inference information. Next, the constructed networks are analyzed to capture the transfer characteristics of fuzzy information granules. The results show that only a few types of fuzzy information granules and fuzzy relation groups play the key role in the fluctuation mechanism, which always have specific targets. Hence, according to the distribution of the transfer probability, a prediction scheme on the granularity level can be established. By utilizing both synthetic and real-life data sets, examples are shown to illustrate the effectiveness and feasibility of the proposed scheme. [ABSTRACT FROM AUTHOR]

Abstract In the study of the long-term prediction of time series with high noise and nonlinearity, the trend information and fluctuation range of the sequence data is usually more valuable and practical than the forecasting of the specific values at time points. A method named as stepwise linear division (SLD) is constructed for the variable-length partition of fuzzy information granule (FIG), characteristics of which are extracted from the original data distribution of time series. Secondly, based on the above partition algorithm, a novel fuzzy information granule is first proposed, which can characterize the variant trend of the data, the range of fluctuation and the degree of dispersion. Moreover, the reliability of the prediction results can be quantified. After granulating time series, a rule-based fuzzy inference system is established to achieve the prediction of time series. Synthetic sequences and real-life time series, including chaotic Mackey–Glass time series, temperature, sunspot numbers, milk production and financial data are utilized in experiments to verify the effectiveness of the proposed scheme. The experimental results demonstrate that the proposed model can contain more rich semantic information and produce better long-term prediction performance compared with existing numeric models and fuzzy inference systems. [ABSTRACT FROM AUTHOR]

In this article, the evolution of prisoner's dilemma game with volunteering on interdependent networks is investigated. Different from the traditional two-strategy game, voluntary participation as an additional strategy is involved in repeated game, that can introduce more complex evolutionary dynamics. And, interdependent networks provide a more generalized network architecture to study the intricate variability of dynamics. We have showed that voluntary participation could effectively promote the density of co-operation, that is also greatly affected by interdependent strength between two coupled networks. We further discussed the influence of interdependent strength on the densities of different strategies and found that an intermediate interdependence would play a bigger role on the evolution of dynamics. Subsequently, the critical values of the defection temptation for phase transitions under different conditions have been studied. Moreover, the global oscillations induced by the circle of dominance of three strategies on interdependent networks have been quantitatively investigated. Counter-intuitively, the oscillations of strategy densities are not periodic or stochastic, but have rich dynamical behaviors. By means of various analysis tools, we have demonstrated the global oscillations of strategy densities possessed chaotic characteristics. [ABSTRACT FROM AUTHOR]

In this article, based on interdependent networks, the effect of self-organized interdependence on the evolution of cooperation is studied. Different from the previous works, the interdependent strength, which can effectively improve the fitness of players, is taken as a kind of limited resources and co-evolves with players’ strategy. We show that the self-organization of interdependent strength would spontaneously lead to power law distribution at the stationary state, where the level of cooperation in system can be significantly promoted. Furthermore, when intermediate quantity of interdependence resources existing in system, the power law distribution is most evident with the power β ≈ 1.72, meanwhile the level of cooperation also reaches the maximum value. We discuss the related microscopic system properties which are responsible for the observed results and also demonstrate that the power law distribution of interdependence resources is an elementary property which is robust against the governing repeated games and the initial resources allocation patterns. [ABSTRACT FROM AUTHOR]

In reality, complicated situations usually include multiple game scenarios, which cannot be effectively described by a single game model. Hence, how to model and reveal the cooperative behaviors in complex multiple game environments is a meaningful topic. In this article, by using switching control, a spatial multigame model with reciprocal reward and reputation mechanism is proposed, where a specific game scenario at each time step is determined by current individuals' behaviors and the game type. Besides, the allocation of limited resources is co-evolved with the evolution of cooperation, which can be regarded as the results of the game and meanwhile affect the while game processes. The evolution dynamics of the proposed model is detailly discussed and the mutual influence of reputation mechanism, reward reciprocity and resource allocation on cooperative behaviors is quantitatively studied. Finally, in order to illustrate the robustness of the analyzed results, its impact on the cooperative behaviors in single PD and SD games are comparatively discussed. • Spatial multigame using switching control. • Reciprocal reward and reputation mechanism in multigame. • Co-evolution of multigame combined with limited resource allocation. [ABSTRACT FROM AUTHOR]

Abstract: In this article, a novel synchronization scheme is proposed to achieve hybrid modified function projective synchronization (HMFPS) in two different dimensional complex nonlinear systems with fully unknown parameters. In the complex space, the response system are asymptotically synchronized up to the different order’s drive system by the state transformation with a scaling function matrix, and all of unknown parameters in both drive and response systems are achieved to be identified. Based on the Lyapunov stability theory, an adaptive controller and updated laws of parameters are developed. Respectively on the ways of increased order and reduced order, the corresponding numerical simulations demonstrate the effectiveness and feasibility of the proposed scheme. [Copyright &y& Elsevier]

• A signal propagation model for financial granular network is constructed. • Theoretical proofs of fluctuation propagation on financial granular network. • Quantitative analysis of the fluctuation propagation in real financial data sets. Financial markets are usually affected by various factors, such as breaking news or the release of important economic data, which would interrupt normal fluctuations and cause the abruptly increasing of volatility. How the financial markets behave under the impact of sudden external stimulations is of great significance for the analysis and prediction of financial prices. In this article, combining the theory of complex networks with granular computing, the financial granular complex network is constructed, based on which the propagation of fluctuation patterns after the occurrence of abnormal changes are studied. Here, the so-called "abnormal changes" are defined as the sudden and fierce fluctuations of financial prices, which are the non-trivial phenomena commonly existing in financial markets. By constructing the dynamical equations of pattern propagation, the evolution trajectories and the intrinsic characteristic of abnormal volatility patterns in financial granular network are quantitatively and qualitatively discussed. Results reveals the behaviors of the financial granule networks after the occurrence of abnormal fluctuations, and the impact of network topology on the propagation of fluctuation patterns is proved theoretically. In order to verify the validity of theoretical analysis, the empirical studies are carried out by using the real financial data deriving from foreign exchange and stock markets. [ABSTRACT FROM AUTHOR]

• A novel spatial multigame model with environment and resources. • Self-organized dynamics from the perspective of resource allocation and environmental factors. • Position-based pinning control mechanism is involved. In reality, environment has the considerable effect on the decision making of individuals. Meanwhile, individuals with different amount of resources could have distinct responses to environmental impacts. In this article, based on spatial multigame model, the mutual effect of environment and resources allocation on cooperation is studied. Firstly, the fitness of individuals in multigame is redefined by using environmental factor and limited resources are introduced into the model coevolving with individual strategy. Secondly, the relationship between resource quantity and strategy selection is studied, where the influence of the environmental factor and resources on cooperative behaviors has been discussed in detail. Thirdly, the evolutionary dynamics of cooperation clusters are discussed such that the characteristics of resource allocation in multigame model are further analyzed. Finally, in order to deeply reveal the impact of resources on the evolution behaviors of the multigame system, position-based pinning control mechanism is involved to control the speed of resource change and the corresponding evolutionary processes are discussed. [ABSTRACT FROM AUTHOR]

In order to reveal the underlying mechanism of aspiration on cooperation with resource allocation, in this article, an aspiration-based game model on interdependent networks is proposed, where the risk-averse voluntary players are introduced into the game model. Each player has a self-adaptive aspiration tuned by the obtained payoff, which will be evolving with cooperation. In reality, limited resources are always related to games. The allocation of resources is generally decided by the results of games and in turn affects the choice of game strategies. Here, limited resources as a co-evolutionary factor is also involved in the proposed model, which exerts influence on the aspiration of players. Based on the prisoner's dilemma, the evolutionary dynamics of the proposed model is discussed in detail. As the results, the self-adaptive aspiration mechanism can significantly improve the level of cooperation. Meanwhile, both the strength of interdependence and the quantity of resources owned by players have great influence on the choice of game strategies. Interestingly, players with higher resources and aspirations are mostly located in the centers of cooperator (loner) strategy clusters. Moreover, based on the proposed game model, public good game and snow game are also implemented to verify the robustness of the results. [ABSTRACT FROM AUTHOR]

In this paper, the co-evolution of cooperation and limited resources in spatial evolutionary game is studied, where memory ability is involved in the evolutionary process. Competition for resources exists widely in nature and human society. Therefore, the allocation of limited resources is always accompanied with various kinds of games. The acquisition of resources depends on the results of the game and the number of resources owned by the players would in turn affect the choice of game strategies. Memory, or so-called experience, usually plays an important role in guiding human behaviors in the real world, which is vital to individual rational decision-making. In order to better simulate the memory ability in real life, memory mechanism with attenuation factor is introduced into the co-evolution of prisoner's dilemma game. Moreover, in order to protect the vulnerable groups, we have introduced the mechanism of minimum guarantee of resources, which simulates the social security system in reality such that the extreme polarization can be alleviated. By using strategy-independent rules, memory ability can effectively raise the level of cooperation, where both the memory length and attenuation factor would affect the co-evolutionary results. Furthermore, limited resources have fast aggregation effect and players with cooperative behaviors have easier access to resources. Meanwhile, the minimum resources guarantee mechanism can considerably affect the cooperation level. Besides, as to the complex dynamical behaviors of co-evolution, quantitative and qualitative discussions are implemented. In order to illustrate the robustness of the obtained results, the cooperative behaviors of systems in public goods game and snow game are also discussed. [ABSTRACT FROM AUTHOR]