Article
Physics, Multidisciplinary
Zhengzhong Yuan, Jingwen Li, Zhesi Shen, Li Hu, Chen Zhao
Summary: Although there is increasing attention on controlling complex networks, an efficient method to analyze the controllability of large undirected networks is still lacking. In this study, a structural reduction method (SRM) is proposed to obtain a control core of the original undirected network by a revised leaf removal process. The SRM demonstrates high efficiency in analyzing the controllability of large undirected networks, and the control core consists of isolated nodes and smaller components, greatly improving the efficiency of identifying and analyzing driver nodes. Furthermore, SRM is applied to empirical networks, confirming its effectiveness by the removal of more than 20% of nodes and 70% of links. These findings deepen our understanding of the controllability of undirected networks and provide an effective method for researching the controllability of large undirected complex networks.
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS
(2023)
Article
Physics, Fluids & Plasmas
Tao Meng, Gaopeng Duan, Aming Li
Summary: Controlling complex networks with reduced energy has been a focus in recent progress. The challenge lies in the high cost of control energy. This study presents a method to exponentially reduce the energy required for controlling complex networks by appropriately setting the initial states of uncontrollable nodes.
Article
Multidisciplinary Sciences
Giacomo Baggio, Danielle S. Bassett, Fabio Pasqualetti
Summary: The paper introduces a data-driven framework for controlling complex networks effectively, even with incomplete or random datasets, which is particularly relevant for power grids and neural networks.
NATURE COMMUNICATIONS
(2021)
Article
Multidisciplinary Sciences
Samie Alizadeh, Marton Posfai, Abdorasoul Ghasemi
Summary: The minimum number of inputs needed for network control is often used to measure its controllability. However, controlling linear dynamics through a minimal set of inputs usually requires excessive energy, and there is a trade-off between reducing the number of inputs and control energy. To study this trade-off, we investigate the problem of identifying a minimum set of input nodes that ensure controllability while limiting the length of the longest control chain. By mapping this problem to a graph combinatorial problem, we show that finding the solution is NP-complete and propose a heuristic approximation. Applying this algorithm to real and model networks, we explore how network structure affects the minimum number of inputs, revealing that for many real networks, reducing the longest control chain only requires rearranging the input nodes, without the need for additional inputs.
SCIENTIFIC REPORTS
(2023)
Article
Computer Science, Artificial Intelligence
Shuai Jia, Yugeng Xi, Dewei Li, Haibin Shao
Summary: This paper introduces a novel algorithmic framework for solving the problem of controlling complex networks, and conducts experiments to analyze the effectiveness and performance.
Article
Engineering, Electrical & Electronic
Guanrong Chen
Summary: This article introduces the notion of pinning control for complex dynamical networks and its application in stabilization, synchronization, and control. It reviews the concept of network pinning control and discusses the fundamental issues of network stabilizability, synchronizability, and controllability. It shows the potential of the self-contained theoretical framework of pinning control technology for practical applications in network science and engineering.
IEEE TRANSACTIONS ON CONSUMER ELECTRONICS
(2022)
Article
Physics, Multidisciplinary
Jie Zhou, Cheng Yuan, Zu-Yu Qian, Bing-Hong Wang, Sen Nie
Summary: The control of complex networks is influenced by their structural characteristic. Cut vertexes, as key nodes in network structure, are crucial for network connectivity. However, their impact on network control remains uncertain. We find that driver nodes tend to avoid cut vertexes, but driving them actually reduces the energy required for controlling complex networks. By investigating the failure of cut vertexes using three different strategies, we show that their failure significantly increases the control energy due to their higher degrees. Our results deepen the understanding of the structural characteristic in network control.
Article
Automation & Control Systems
Baike She, Siddhartha Mehta, Chau Ton, Zhen Kan
Summary: This article investigates energy-related controllability of complex networks, focusing on the energy incurred by leaders in controlling signed networks. By utilizing controllability Gramian-based measures and graph Laplacian analysis, fundamental relationships between control difficulty and network topology are discovered.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2021)
Review
Automation & Control Systems
Giacomo Baggio, Fabio Pasqualetti, Sandro Zampieri
Summary: This article introduces the fundamental principles and limitations of controlling complex networks, presents an energy-aware controllability metric, discusses its properties and bounds, and examines the problem of optimally selecting a set of control nodes to minimize control effort.
ANNUAL REVIEW OF CONTROL ROBOTICS AND AUTONOMOUS SYSTEMS
(2022)
Article
Mechanics
Zu-Yu Qian, Cheng Yuan, Jie Zhou, Shi-Ming Chen, Sen Nie
Summary: This study explores the incorporation of conformity behavior into network control and finds that controlling undirected networked systems with conformity becomes easier after the network connectivity reaches a critical point. The research also identifies key nodal structural characteristics and proposes an optimal control strategy to reduce energy consumption. These findings are validated in synthetic and real networks, highlighting their significance in describing control energy in networked systems.
JOURNAL OF STATISTICAL MECHANICS-THEORY AND EXPERIMENT
(2022)
Article
Multidisciplinary Sciences
Yuma Shinzawa, Tatsuya Akutsu, Jose C. Nacher
Summary: The study introduces a new constraint where one driver node can only control one target node and develops an algorithm to classify the driven nodes. Computational analysis shows that the number of driven nodes is significantly larger than the number of driver nodes, indicating their importance in real-world network analysis.
SCIENTIFIC REPORTS
(2021)
Article
Automation & Control Systems
Dongsheng Yang, Yunhe Sun, Qinglai Wei, Huaguang Zhang, Ting Li
Summary: This article proposes a topology predicting method based on the connection probability matrix for analyzing the structural controllability of complex networks without connection information, and establishes modified principles for predicting global network topologies and determining the drive node set of networks. The accuracy of this method is verified through numerical simulations, showing that it can accurately predict the global topology and structural controllability of complex networks with large scale and high edge density.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2022)
Article
Engineering, Industrial
Lin Zhang, Huiying Wen, Jian Lu, Da Lei, Shubin Li, Satish Ukkusuri
Summary: This study proposes a cascading reliability model for multi-modal public transit networks to address the issue of cascading failures. By considering the impact of other transit modes networks, the model enables efficient testing and calculation of network reliability, and its adaptability and controllability are validated through a case study.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2022)
Article
Automation & Control Systems
Lei Chen, Xinghuo Yu, Xin Xin, Changyin Sun
Summary: This paper studies characteristic model-based modeling and control approaches for complex dynamical networks based on sampled data. The characteristic model simplifies the underlying network's topological structures to provide a straightforward and implicit description for network dynamics, and the induced parameter estimation method makes the model adaptive and purely data-driven. A control law based on this model is proposed to govern the network dynamics, and the theoretical results are verified through numerical simulations.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Automation & Control Systems
Paulo Tabuada, Bahman Gharesifard
Summary: In this article, it is demonstrated that deep residual neural networks can be modeled as nonlinear control systems and have the capability of universal approximation. The problem of exact memorization of training data using a deep residual neural network is studied by formulating it as a controllability problem. Through geometric control theory, a class of activation functions is identified to ensure controllability on an open and dense submanifold of sample points. It is further proven that any continuous function on a compact set can be approximated to arbitrary accuracy (with respect to the uniform norm) by this class of neural networks, with optimal bounds on the number of required neurons.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2023)
Article
Physics, Multidisciplinary
Yanchen Liu, Nima Dehmamy, Albert-Laszlo Barabasi
Summary: The study introduces the concept of network isotopy and graph linking number to distinguish physical networks with identical wiring but different layouts. It reveals that each local tangle contributes independently to the total energy of the network, aiding in the formulation of a statistical model. Application of this framework shows that the entanglement level in the mouse connectome exceeds expectations based on optimal wiring.
Article
Physics, Multidisciplinary
Ling Zhan, Tao Jia
Summary: This article highlights the importance of Heterogeneous Information Network (HIN) embedding and proposes a new embedding method called CoarSAS2hvec. The method incorporates self-avoiding short sequence sampling and an optimized loss function to improve the performance of HIN structure embedding. CoarSAS2hvec outperforms other methods in node classification and community detection tasks, and effectively captures richer information compared to existing methods.
Article
Mathematics, Applied
Radoslaw Michalski, Damian Serwata, Mateusz Nurek, Boleslaw K. Szymanski, Przemyslaw Kazienko, Tao Jia
Summary: This research develops a temporal network epistemology model to simulate the learning process in dynamic networks. The results demonstrate that the temporal dynamics of the network significantly influence the outcome and flow of the learning process, leading to different consensus dynamics and previously unobserved phenomena. The approach and experimental results provide insights into how human communities collectively solve complex problems and spread beliefs across societies.
Article
Computer Science, Artificial Intelligence
Yansong Wang, Xiaomeng Wang, Yijun Ran, Radoslaw Michalski, Tao Jia
Summary: One important task in the study of information cascade is to predict the future recipients of a message given its past spreading trajectory. The proposed method CasSeqGCN combines the network structure and temporal feature to accurately predict the future cascade size. The experiment demonstrates that the improved prediction comes from the design of the input and the GCN layer.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Psychiatry
Yunsong Luo, Wenyu Chen, Jiang Qiu, Tao Jia
Summary: This study used a large-scale resting-state fMRI dataset from China to develop a stacking machine learning model that estimates the brain aging of MDD patients. The results showed that MDD patients have a higher brain-predicted age difference compared to healthy controls, and antidepressant users in the MDD subgroup have an even greater age difference. This study confirms the presence of accelerated brain aging in MDD patients and verifies existing findings using functional brain connectivity.
TRANSLATIONAL PSYCHIATRY
(2022)
Article
Multidisciplinary Sciences
Tunde Pacza, Mayara L. Martins, Maha Rockaya, Katalin Muller, Ayan Chatterjee, Albert-Laszlo Barabasi, Jozsef Baranyi
Summary: This study develops a database called MilkyBase that contains the biochemical composition of human milk. The data are selected, digitized, and curated using both machine-learning and manual methods. The database allows users to find patterns in milk composition based on various factors and provides a platform for users to input their own data. The database is user-friendly and facilitates statistical analysis, uncertainty quantification, and prediction of dynamic compositions.
Article
Automation & Control Systems
Tianyang Cai, Tao Jia, Sridhar Adepu, Yuqi Li, Zheng Yang
Summary: With the widespread innovation of IoT, SDN, and cloud computing, cyber-physical systems (CPSs) have been developed to facilitate daily life and economy. However, the shutdown of critical CPSs and the increasing threat of DDoS attacks pose serious consequences. This article presents an adaptive DDoS attack mitigation scheme called ADAM, which combines information entropy and unsupervised anomaly detection methods to accurately detect and mitigate DDoS attacks. Experimental results show that ADAM has a high accuracy in mitigating DDoS attacks and reduces the false-positive rate compared to similar work.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Multidisciplinary Sciences
Hao Wu, Xiangyi Meng, Michael M. Danziger, Sean P. Cornelius, Hui Tian, Albert-Laszlo Barabasi
Summary: The understanding of recovery processes in power distribution grids is limited by the lack of realistic outage data, especially large-scale blackout datasets. By analyzing data from three electrical companies across the United States, researchers found that the recovery duration of an outage is connected with the downtime of its nearby outages and blackout intensity, but is independent of the number of customers affected. They presented a cluster-based recovery framework to analyze the dependence between outages and interpret the dominant role blackout intensity plays in recovery.
NATURE COMMUNICATIONS
(2022)
Article
Multidisciplinary Sciences
Csaba Both, Nima Dehmamy, Rose Yu, Albert-Laszlo Barabasi
Summary: Graph layout algorithms are crucial in visualizing complex networks, but the current force-directed layout (FDL) algorithm has limitations in terms of computational complexity and layout interpretability. This study proposes using Graph Neural Networks (GNN) to accelerate FDL, resulting in significant speed improvement and more informative layouts. The use of deep learning in network visualization can also have implications for other optimization problems.
NATURE COMMUNICATIONS
(2023)
Article
Multidisciplinary Sciences
Giulia Menichetti, Babak Ravandi, Dariush Mozaffarian, Albert-Laszlo Barabasi
Summary: This study introduces a machine learning algorithm that accurately predicts the degree of processing for any food, indicating that over 73% of the US food supply is ultra-processed. The research shows that increased reliance on ultra-processed food is associated with various health issues and replacing these foods with less processed alternatives can significantly reduce their negative health implications. Thus, providing information on the degree of processing could improve population health.
NATURE COMMUNICATIONS
(2023)
Article
Multidisciplinary Sciences
Deisy Morselli Gysi, Albert-Laszlo Barabasi
Summary: Combining interactions mediated by noncoding RNAs (ncRNAs) with protein-protein interactions (PPI) improves the identification of disease modules and reveals previously undetectable disease relationships and comorbidity patterns.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2023)
Article
Information Science & Library Science
Yang Li, Tao Jia
Summary: The purpose of this study is to propose an improved credit allocation method that enhances the distinguishability of the leading author and the robustness to malicious manipulations. By using a modified Sigmoid function and excluding the target paper in calculating co-citation contributions, the proposed method (NCCAS) is tested on Nobel Prize-winning papers and a large dataset of computer science articles. The findings indicate that NCCAS provides the most accurate prediction of Nobel laureates, ensures the distinguishability of the leading author, and exhibits robustness against malicious manipulations.
JOURNAL OF DATA AND INFORMATION SCIENCE
(2023)
Article
Food Science & Technology
Giulia Menichetti, Albert-Laszlo Barabasi
Summary: This study quantifies the nutrient concentrations in food using a mathematical model and proposes a method to fill in missing values in food composition databases. The research shows that the concentration of each nutrient in food follows a universal scaling law, which can be explained by biochemical constraints. The findings provide a mathematical basis for understanding the impact of food processing on nutrient balance and health effects.
Article
Mathematics, Interdisciplinary Applications
Yijun Ran, Si-Yuan Liu, Xiaoyao Yu, Ke-Ke Shang, Tao Jia
Summary: This article discusses the importance of link prediction in evolving systems and proposes a new method to address the temporal link prediction problem with new nodes.
JOURNAL OF PHYSICS-COMPLEXITY
(2022)
Article
Information Science & Library Science
Linlin Liu, Jianfei Yu, Junming Huang, Feng Xia, Tao Jia
Summary: A recent study shows that research in China is more dominated by large teams, particularly in the field of natural science. Despite the global trend of more papers being written by big teams, China has seen a significant drop in output from small teams. This imbalance highlights the need to strike a balance between small and big teams for fostering innovation and originality.
QUANTITATIVE SCIENCE STUDIES
(2021)