State Of Complex System Research Articles

Estimating reaction barriers with deep reinforcement learning1

Stable states in complex systems correspond to local minima on the associated potential energy surface. Transitions between these local minima govern the dynamics of such systems. Precisely determining the transition pathways in complex and high-dimensional systems is challenging because these transitions are rare events, and isolating the relevant species in experiments is difficult. Most of the time, the system remains near a local minimum, with rare, large fluctuations leading to transitions between minima. The probability of such transitions decreases exponentially with the height of the energy barrier, making the system’s dynamics highly sensitive to the calculated energy barriers. This work aims to formulate the problem of finding the minimum energy barrier between two stable states in the system’s state space as a cost-minimization problem. It is proposed to solve this problem using reinforcement learning algorithms. The exploratory nature of reinforcement learning agents enables efficient sampling and determination of the minimum energy barrier for transitions.

A health assessment method with attribute importance modeling for complex systems using belief rule base

In the health assessment for complex systems, system attributes refer to the indicators or components having an impact on the health state of the system. When assessing the health state of complex systems, the relative importance of system attributes should be accurately modeled, otherwise incorrect results may occur. Especially for some complex systems with small samples, data-driven methods are prone to overfitting. In this article, a new method using belief rule base (BRB) is proposed to model the relative importance of system attributes while assessing the health state. As an interpretable modeling tool, BRB constructs nonlinear mapping relationships between system attributes and health state. A data-knowledge hybrid-driven ensemble feature selection (DKH-EFS) method is developed to calculate the relative importance of the system attributes, namely attribute importance (AM). A random sensitivity analysis (RSA) method of the model output to the input of BRB is proposed to calculate the feature importance (FI) of BRB. A new parameter optimization model considering the consistency between the AM and FI is introduced to improve the modeling ability of BRB for the AM. A case study on the health assessment of the fiber optic gyroscope (FOG) validates the proposed method.

Cross-Recurrence Quantification Of Cardiovascular Signals In Newborns Is A Sensitive Marker Of Health Status

Background — Analysis of the state of complex biological systems requires the use of sensitive methods for diagnosing interactions using experimental time series. Objective — To evaluate the possibility of using the technique of cross-recurrence quantification (CRQ) in studying the health status of newborns using low-frequency components of RR-interval signals and photoplethysmograms, reflecting the dynamics of the circuits in autonomic regulation of blood circulation. Methods — The study included two groups of neonates: 10 full-term newborns and 10 preterm neonates. We carried out simultaneous recording of electrocardiographic and photoplethysmographic signals. CRQ analysis was employed as the primary tool. Results — We established that some indices of CRQ analysis, characterizing the degree of interaction of the studied circuits, act as sensitive markers. They make it possible to distinguish the dynamics of the studied contours between healthy newborns and preterm neonates. Conclusion — The results of our study confirmed that CRQ is a promising tool in creating methods for diagnosing health conditions, including in newborns.

Application of the neural network apparatus for solving tasks of forecasting the state of complex systems.

Application of the neural network apparatus for solving tasks of forecasting the state of complex systems.

Major Role of Multiscale Entropy Evolution in Complex Systems and Data Science.

Complex systems are prevalent in various disciplines encompassing the natural and social sciences, such as physics, biology, economics, and sociology. Leveraging data science techniques, particularly those rooted in artificial intelligence and machine learning, offers a promising avenue for comprehending the intricacies of complex systems without necessitating detailed knowledge of underlying dynamics. In this paper, we demonstrate that multiscale entropy (MSE) is pivotal in describing the steady state of complex systems. Introducing the multiscale entropy dynamics (MED) methodology, we provide a framework for dissecting system dynamics and uncovering the driving forces behind their evolution. Our investigation reveals that the MED methodology facilitates the expression of complex system dynamics through a Generalized Nonlinear Schrödinger Equation (GNSE) that thus demonstrates its potential applicability across diverse complex systems. By elucidating the entropic underpinnings of complexity, our study paves the way for a deeper understanding of dynamic phenomena. It offers insights into the behavior of complex systems across various domains.

Research on Online Condition monitoring for Complex System based on Modified Broad Learning Systems

Complex systems contain numerous interacting components, thus deep learning methods with powerful performance and complex structure are often used to achieve condition monitoring. However, the deep learning methods are always too time-consuming and hardware-demanding to be loaded into complex systems for online training and updates. To achieve accurate and timely monitoring of complex system state, based on broad learning system (BLS), an online condition monitoring method is proposed in this paper. General BLSs are based on a randomly generated hidden-layer, usually perform poorly in high-dimensional data classification tasks. In this work, based on correlation and causality, two modified BLSs are proposed and mixed to establish the online monitoring system. Specifically, logistic regression (LR) and structural causal model (SCM) are considered to form rough predictions of the system state, thus to replace the randomly generated ones with no practical significance. The effectiveness of the proposed online monitoring method is verified by both simulation data and real data.

MIWE: detecting the critical states of complex biological systems by the mutual information weighted entropy

Complex biological systems often undergo sudden qualitative changes during their dynamic evolution. These critical transitions are typically characterized by a catastrophic progression of the system. Identifying the critical point is critical to uncovering the underlying mechanisms of complex biological systems. However, the system may exhibit minimal changes in its state until the critical point is reached, and in the face of high throughput and strong noise data, traditional biomarkers may not be effective in distinguishing the critical state. In this study, we propose a novel approach, mutual information weighted entropy (MIWE), which uses mutual information between genes to build networks and identifies critical states by quantifying molecular dynamic differences at each stage through weighted differential entropy. The method is applied to one numerical simulation dataset and four real datasets, including bulk and single-cell expression datasets. The critical states of the system can be recognized and the robustness of MIWE method is verified by numerical simulation under the influence of different noises. Moreover, we identify two key transcription factors (TFs), CREB1 and CREB3, that regulate downstream signaling genes to coordinate cell fate commitment. The dark genes in the single-cell expression datasets are mined to reveal the potential pathway regulation mechanism.

Development of a method for increasing the efficiency of decision-making in special purpose systems

The object of the research is special purpose systems.
 Investigated problem: The experience of operations (combat operations) of recent years shows the growing role of information influence measures on the systems of collection, processing and transmission of special purpose information and decision-making officials.
 The main scientific results: The specificity of measures to ensure the information security of the state is that, on the one hand, it is necessary to solve the task of collecting, processing and transmitting information, and on the other hand, it is necessary to counteract measures of information influence on the systems of collecting, processing and transmitting information and decision-making officials.
 Victory over a quantitatively superior opponent is possible only if there is a technological advantage over it. One of such technological approaches is the use by the defense forces of Ukraine of the concept of the Internet of Things (IoT)
 The essence of this concept involves the use of a large number of sensors, sensors, unmanned aerial vehicles, as well as other technical means (tablets, communication means) combined into a single network, which are united by a common goal of functioning.
 The area of practical use of the research results: It is advisable to use the proposed scientific results when conducting research and development works on the creation of intelligent systems for collecting, processing and analyzing information about the special purpose systems, and developing requirements for hardware and software of this type of systems.
 Innovative technological product: the proposed method is recommended to be used as part of the software of special-purpose systems. This method is advisable to apply at the stage of assessing the state of special-purpose systems and further making managerial decisions by the persons taking them. Also, the proposed method should be incorporated as an integral part of decision support systems at the stage of analyzing the state of complex technical systems.
 Field of application: software, special purpose systems, decision support systems.

The development of the method of evaluation of complex hierarchical systems based on improved alforitm of particle swarm

The scientific task, which is solved in the research, is to increase the efficiency of the evaluation of complex hierarchical real-time systems. Finding solutions to nonlinear optimization problems and especially global optimization problems is one of the most popular problems in computational mathematics. In applied problems, the objective function, as a rule, has a large number of variables, is not given in an analytical form and is calculated as some integral characteristic of a complex dynamic process. The development of effective methods, to a certain extent adaptive to the variability of the objective function, is especially relevant in connection with the development of computer technology and the possibility of using parallel computing systems. The conducted research was aimed at developing a method of evaluating complex hierarchical systems based on an improved particle swarm. At the same time, the object of research was complex hierarchical real-time systems. The subject of research is the functioning of real-time hierarchical systems. The novelties of the proposed method consist in: ‒ creating a multi-level and interconnected description of complex systems of hierarchical real-time systems; ‒ increasing the efficiency of decision making while evaluating complex systems of hierarchical real-time systems; ‒ solving the problem of falling into global and local extremes while assessing the state of complex systems of hierarchical real-time systems; ‒ the possibilities of directed search by several individuals of the particles swarm in a given direction, taking into account the degree of uncertainty; ‒ the possibilities of re-analysis of the state of complex systems of hierarchical real-time systems; ‒ avoiding the problem of loops while visualizing the state of the national security system in real time. It is advisable to implement the specified method in specialized software, which is used to analyze the state of complex systems of hierarchical real-time systems and make management decisions.

METHOD OF ASSESSING THE STATE OF HIERARCHICAL OBJECTS BASED ON BIO-INSPIRED ALGORITHMS

Relevance. Nowadays, no state in the world is able to work on the creation and implementation of artificial intelligence in isolation from others. Artificial intelligence technologies are actively used to solve both general and highly specialized tasks in various spheres of society. In the process of assessing (identifying) the state of complex and objects of analysis and management, there is a high degree of a priori uncertainty regarding their state and a small amount of initial data describing them. At the same time, despite the huge amount of information, the degree of non-linearity, illogicality and noisy data is increasing. That is why the issue of improving the efficiency of assessing the condition of complex and objects is an important and urgent issue. The object of research is the objects of analysis. The subject of the research is the identification and forecasting of the analysis objects state with the help of bio-inspired algorithms. In the research, the evaluation and forecasting method was developed using fuzzy cognitive maps and the genetic algorithm. The novelty of the proposed method consists in: taking into account the degree of uncertainty about the object state while calculating the correction factor; adding a correction factor for data noise as a result of distortion of information about the object state; reduction of computing costs while assessing the objects state; creation of a multi-level and interconnected description of hierarchical objects; adjusting the description of the object as a result of changing its current state using a genetic algorithm; the possibility of performing calculations with the original data, which are different in nature and units of measurement. It is advisable to implement the mentioned method in specialized software, which is used to analyze the state of complex technical systems and make decisions.

Transition between metastable equilibria: Applications to binary-choice games.

Transitions between metastable equilibria in the low-temperature phase of dynamical Ising game with activity spillover are studied in the infinite time limit. It is shown that exponential enhancement due to activity spillover, which takes place in finite-time transitions, is absent in the infinite time limit. In order to demonstrate that, the analytical description for infinite time trajectory is developed. An analytical approach to estimate the probability of transition between metastable equilibria in the infinite time limit is introduced and its results are compared with those of kinetic Monte Carlo simulation. Our study sheds light on the dynamics of the Ising game and has implications for the understanding of transitions between metastable states in complex systems.

Measuring the Ultrafast Correlation Dynamics of a Multilevel System Using the Center Line Slope Analysis in Two-Dimensional Electronic Spectroscopy.

In a two-dimensional (2D) optical spectrum of a multilevel system, there are diagonal peaks and off-diagonal cross-peaks that correlate the different levels. The time-dependent properties of these diagonal peaks and cross-peaks contain much information about the dynamics of the multilevel system. The time-dependent diagonal peakshape that depends on the spectral diffusion dynamics of the associated transition and characterized by the frequency-fluctuation correlation function (FFCF) is well studied. However, the time-dependent peakshape of a cross-peak that provides the correlation dynamics between different transitions is much less studied or understood. We derived the third-order nonlinear response functions that describe the cross-peaks in a 2D electronic spectrum of a multilevel system that arise from processes sharing a common ground state and/or from internal conversion and population transfer. We can use the center line slope (CLS) analysis to characterize the cross-peaks in conjunction with the diagonal peaks. This allows us to recover the frequency-fluctuation cross-correlation functions (FXCFs) between two transitions. The FXCF and its subsidiary quantities such as the initial correlation and the initial covariance between different transitions are important for studying the correlation effects between states in complex systems, such as energy-transfer processes. Furthermore, knowledge of how various molecular processes over different timescales affect simultaneously different transitions can also be obtained from the measured FXCF. We validated and tested our derived equations and analysis process by studying, as an example, the 2D electronic spectra of metal-free phthalocyanine in solution. We measured and analyzed the diagonal peaks of the Qx and Qy transitions and the cross-peaks between these two transitions of this multilevel electronic system and obtained the associated FFCFs and FXCFs. In this model system, we measured negative components of FXCF over the tens of picosecond timescale. This suggests that in phthalocyanine, the Qx and Qy transitions coupling with the solvent molecule motion are anticorrelated to each other.

The development of a method for assessing the security of complex technical systems using artificial immune systems

Ensuring the security of complex technical systems of various functional purposes requires a constant search for new scientific and practical approaches in order to ensure its proper level against a growing list of new risks and threats. Nowadays, no state in the world is able to work on the creation and implementation of artificial intelligence in isolation from others. Artificial intelligence technologies are actively used to solve both general and highly specialized tasks in various spheres of society. The problem of synthesis of management of complex technological processes is an urgent task in management theory. A promising direction in the design of such complex ones is the use of bio-inspired algorithms that are effectively used while solving optimization tasks. Thus, the object of research is complex technical systems. The subject of research is the state security of complex technical systems. The research developed a method for assessing the security of complex technical systems using artificial immune systems. The novelty of the proposed method consists in: ‒ taking into account while calculating the correction factor for the degree of uncertainty about the state of a complex technical system; ‒ reducing computing costs while assessing the state of a complex technical system; ‒ improved implementation of procedures for solving the task of influencing relationships in a complex technical system; ‒ creating a multi-level and interconnected description of hierarchical complex technical systems; ‒ the possibility of performing calculations with source data that are different in nature and units of measurement. It is advisable to implement the mentioned technique in specialized software, which is used to analyze the state of complex technical systems and make decisions.

Interval type-2 generalizing fuzzy model for monitoring the states of complex systems using expert knowledge

A type-2 interval generalizing fuzzy model is proposed for monitoring complex systems’ states. A set-theoretic approach is proposed to generalize the results of type-2 fuzzy models with interval membership functions. The study of the correspondence of expert assessments to the output value of the generalizing fuzzy model over the observation interval is presented. Examples of the use of generalizing fuzzy model type-2 in the task of monitoring the conditions of an artesian well are given. It is shown that in order to improve the quality of decisions made, the expert needs to pay attention to the value of the interval output of the generalizing type-2 fuzzy model. Recommendations are presented to experts to improve decision-making regarding the estimation of the output interval of the generalizing model.

Development of assessment and forecasting techniques using fuzzy cognitive maps

Nowadays, no state in the world is able to work on the creation and implementation of artificial intelligence (AI) in isolation from others. AI technologies are used to solve general and highly specialized tasks in various spheres of society. In the process of assessing (identifying) the state of complex objects and objects of management analysis, there is a high degree of a priori uncertainty regarding their state and a small amount of initial data describing them. At the same time, despite the huge amount of information, the degree of non-linearity, illogicality and noisy data is increasing. That is why the issue of improving the efficiency of assessing the condition of components and objects is an important issue. Thus, the objects of analysis were chosen as the research object. The subject of research is the identification and forecasting of the analysis object. In the research, the evaluation and forecasting method was developed using fuzzy cognitive maps. The features of the proposed method are: ‒ taking into account the degree of uncertainty about the object state while calculating the correction factor; ‒ adding a correction factor for data noise as a result of distortion of information about the object state; ‒ reduction of computing costs while assessing the object state; ‒ creation of a multi-level and interconnected description of hierarchical objects; ‒ correction of the description of the object as a result of a change in its current state using a genetic algorithm; ‒ the possibility of performing calculations with source data that are different in nature and units of measurement. It is advisable to implement the proposed method in specialized software, which is used to analyze the state of complex technical systems and while making decisions.

Dynamical Analysis of the Incommensurate Fractional-Order Hopfield Neural Network System and Its Digital Circuit Realization

Dynamical analysis of the incommensurate fractional-order neural network is a novel topic in the field of chaos research. This article investigates a Hopfield neural network (HNN) system in view of incommensurate fractional orders. Using the Adomian decomposition method (ADM) algorithm, the solution of the incommensurate fractional-order Hopfield neural network (FOHNN) system is solved. The equilibrium point of the system is discussed, and the dissipative characteristics are verified and discussed. By varying the order values of the proposed system, different dynamical behaviors of the incommensurate FOHNN system are explored and discussed via bifurcation diagrams, the Lyapunov exponent spectrum, complexity, etc. Finally, using the DSP platform to implement the system, the results are in good agreement with those of the simulation. The actual results indicate that the system shows many complex and interesting phenomena, such as attractor coexistence and an inversion property, with dynamic changes of the order of q0, q1, and q2. These phenomena provide important insights for simulating complex neural system states in pathological conditions and provide the theoretical basis for the later study of incommensurate fractional-order neural network systems.

Deep attention fuzzy cognitive maps for interpretable multivariate time series prediction

Although time series prediction is widely used to estimate the future state of complex systems in various industries, accurate, interpretable and generalizable methods are still limited when used to make long-term nonstationary predictions. To this end, this article proposes deep attention fuzzy cognitive maps (DAFCM), which is composed of spatiotemporal fuzzy cognitive maps (STFCM), long short-term memory (LSTM) neural network, temporal fuzzy cognitive maps (TFCM) and residual structures. First, an improved attention mechanism is used to build spatiotemporal fuzzy cognitive maps that capture the spatial correlation in pairs of nodes and the temporal correlation of respective nodes. Second, the node state updated through the STFCM is input to the LSTM to capture the long-term trend of these series, and the TFCM with improved time attention is applied for the nonstationary problem in the time series. Finally, we add the state values of previous nodes into the DAFCM and build residual structures through linear transformation to prevent gradient explosion and gradient disappearance in long-term backpropagation. By combining the interpretability of fuzzy cognitive maps (FCM) and the high prediction accuracy of deep learning, the DAFCM can be used to accomplish tasks such as multivariate long-term nonstationary time series forecasting in multiple domains, and its efficiency is validated with 6 public datasets across 9 baselines.

Development of methods for identifying the state of various dynamic objects

Artificial intelligence technologies are actively used to solve both general and highly specialized tasks. In the process of assessing (identifying) the condition of complex and heterogeneous objects, there is a high degree of a priori uncertainty regarding their condition and a small amount of initial data describing them. The trends of armed conflicts of the last decades and the regularities of the development of information systems, convincingly indicate the need to change approaches to the collection of information from various sources and their analysis. There is a constant transformation of the forms of information presentation and the order of storage and access to various types of data. The problem of integrating disparate sources of information collection into a single information space is also not fully resolved. That is why the issue of improving the efficiency of assessing the state of complex and heterogeneous dynamic objects is an important and urgent issue. The objects of research are heterogeneous dynamic objects. The subject of the research is the identification of the state of heterogeneous dynamic objects. In the research, the method of identifying the state of heterogeneous dynamic objects was developed. The novelty of the proposed method consists in: ‒ taking into account the degree of uncertainty about the state of a heterogeneous dynamic object; ‒ taking into account the degree of data noise as a result of distortion of data characterizing the state of a heterogeneous dynamic object; ‒ reducing computing costs while assessing the state of heterogeneous dynamic objects; ‒ the possibility of performing calculations with source data that are different in nature and units of measurement. It is advisable to implement the mentioned method in specialized software, which is used to analyze the state of complex technical systems and make decisions.

Resource Allocation With Workload-Time Windows for Cloud-Based Software Services: A Deep Reinforcement Learning Approach

As the workloads and service requests in cloud computing environments change constantly, cloud-based software services need to adaptively allocate resources for ensuring the Quality-of-Service (QoS) while reducing resource costs. However, it is very challenging to achieve adaptive resource allocation for cloud-based software services with complex and variable system states. Most of the existing methods only consider the current condition of workloads, and thus cannot well adapt to real-world cloud environments subject to fluctuating workloads. To address this challenge, we propose a novel Deep Reinforcement learning based resource Allocation method with workload-time Windows (DRAW) for cloud-based software services that considers both the current and future workloads in the resource allocation process. Specifically, an original Deep Q-Network (DQN) based prediction model of management operations is trained based on workload-time windows, which can be used to predict appropriate management operations under different system states. Next, a new feedback-control mechanism is designed to construct the objective resource allocation plan under the current system state through iterative execution of management operations. Extensive simulation results demonstrate that the prediction accuracy of management operations generated by the proposed DRAW method can reach 90.69%. Moreover, the DRAW can achieve the optimal/near-optimal performance and outperform other classic methods by 3~13% under different scenarios.

Frequency-domain physical constrained neural network for nonlinear system dynamic prediction

The current use of physical constraints as prior knowledge of neural network models is a new frontier for artificial intelligence to solve complex nonlinear system predictions. Aiming at the problems of limited feature expression and low time-domain computational efficiency of existing physical-constrained neural networks, a physical informed neural network (PINN) modeling approach in frequency domain for nonlinear system dynamic prediction is proposed. First, the frequency-domain physics-constrained neural network (FPNN) framework is constructed, which mainly consists of CliqueNet and Fourier spectral method (FSM). Then, the CliqueNet with convolution feature cyclic structure for high dimensional sequential data is designed. There are forward and backward connections between any two layers of the network to improve the cross-level information flow of the deep network. Finally, FSM is utilized to inform the neural network of physical states by learning specific initial conditions without training data. Simulation results show that FPNN can effectively achieve Kuramoto–Sivashinsky complex system state prediction, with higher computational efficiency and faster learning efficiency compared with SOTA approaches.