Article
Computer Science, Information Systems
Jie Chen, Nan Song, Yansen Su, Shu Zhao, Yanping Zhang
Summary: This study designed a novel approach to analyze the sentiment orientation of users in social networks, improving sentiment analysis performance by integrating user interactions and opinion data.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Oindrila Chatterjee, Shantanu Chakrabartty
Summary: This article introduces a new energy-efficient learning framework that leverages structural and functional similarities between machine-learning and electrical networks. Unlike traditional energy-based learning models, this framework associates active and reactive energy, dissipating only active power during the learning process.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Multidisciplinary Sciences
Chengyi Tu, Paolo D'Odorico, Samir Suweis
Summary: One of the main challenges in complexity science and engineering is understanding high-dimensional networked systems and their susceptibility to transitions to undesired states. The authors propose an analytical framework to collapse complex N-dimensional networked systems into a lower-dimensional manifold, which can simplify the study of system dynamics and help in identifying optimal strategies in the design or management of networked systems.
Article
Mathematics, Applied
Nicholas W. Landry, Juan G. Restrepo
Summary: This article discusses the concept of the largest eigenvalue of a matrix in hypergraphs and introduces the expansion eigenvalue for dynamical processes in hypergraphs. It provides approximations to the expansion eigenvalue using different methods and explores the application of the expansion eigenvalue in assortative hypergraphs. The article also suggests how reducing dynamical assortativity in hypergraphs can help to extinguish epidemics.
Article
Neurosciences
Jennifer Rollo, John Crawford, John Hardy
Summary: Alzheimer's disease is a complex pathology that requires non-linear dynamical systems modeling and community-wide participation to understand and predict its dynamics. A new methodology is proposed to integrate intuition and test system-level hypotheses and interventions.
Article
Business, Finance
Jin Shao, Jingke Hong, Xianzhu Wang, Xiaochen Yan
Summary: This study investigates the dynamic relationship between house prices and social media sentiment in China. A housing sentiment index is created using natural language processing techniques, and wavelet analysis is used to examine causal correlations. The findings show that the sentiment index is negatively correlated with house price fluctuations overall. The long-term relationship between house prices and sentiment is bidirectional, while house prices causally affect sentiment in the short term. Additionally, sentiment significantly impacts house prices in third-tier cities and the western regions.
FINANCE RESEARCH LETTERS
(2023)
Article
Mathematics
Diana Ogorelova, Felix Sadyrbaev, Inna Samuilik
Summary: This article discusses the issue of targeted control over trajectories of systems of differential equations encountered in genetic and neural network theories. It provides examples of transferring trajectories from the basin of attraction of one attractor to the basin of attraction of a target attractor. The article focuses on a system of ordinary differential equations in gene network theory, where each trajectory describes the current and future states of the network. The possibility of reorienting a given trajectory from the initial state to an assigned attractor is explored, which implies partial control of the network. The difficulty lies in selecting parameters that lead to the desired outcome. Similar problems arise when modeling the response of the body's gene networks to serious diseases such as leukemia. Solving these problems is the first step in applying mathematical methods in medicine and pharmacology.
Article
Biochemical Research Methods
Elena S. Dimitrova, Adam C. Knapp, Brandilyn Stigler, Michael E. Stillman
Summary: This study proposes Cyclone, a software package that can simulate the complete state space of a finite dynamical system over any finite set. Unlike other software packages, Cyclone is capable of computing the complete state space of function-based models.
Article
Mathematics, Interdisciplinary Applications
Chengyi Tu, Jianhong Luo, Ying Fan, Xuwei Pan
Summary: Dimensionality reduction is a powerful tool for analyzing complex systems and uncovering their underlying mechanisms and phenomena. We have developed a framework for dimensionality reduction of stochastic complex dynamical networks, which can capture the essential features and long-term dynamics of the original system in a low-dimensional effective equation.
CHAOS SOLITONS & FRACTALS
(2023)
Article
Automation & Control Systems
Stefanie Winkler, Andreas Koerner, Felix Breitenecker
Summary: The text describes the use of hybrid approaches in various industry branches to solve complex application problems. Different tools for simulation and identification of hybrid systems have been developed over the last decades. The integration of artificial feed-forward neural networks into the modelling process of HDS allows for interdisciplinary exchange and introduces specific modelling methods and challenges.
NONLINEAR ANALYSIS-HYBRID SYSTEMS
(2021)
Article
Biodiversity Conservation
Mingkang Yuan, Yi Xiao, Yang Yang, Cong Liu
Summary: Post-disaster reconstruction is crucial for achieving sustainable development in affected regions. This study explores the coupling and coordinated development of the economic-ecological-social (EES) complex system during the post-disaster reconstruction following the Wenchuan earthquake in China. The results indicate that post-disaster reconstruction significantly accelerates sustainable development, with distinct geographic stratification differences. Factors such as geographic location, ecological resource endowment, science and education, and the labor force are closely related to the coordination degree of the complex system.
ECOLOGICAL INDICATORS
(2023)
Article
Multidisciplinary Sciences
Daniel Fernex, Bernd R. Noack, Richard Semaan
Summary: This study introduces a universal method for data-driven modeling of complex nonlinear dynamics, bridging machine learning, network science, and statistical physics. The proposed cluster-based network modeling (CNM) describes short- and long-term behavior and is fully automatable. This approach complements network connectivity science and offers fast-track avenues for understanding, estimating, predicting, and controlling complex systems in all scientific fields.
Article
Mathematics, Interdisciplinary Applications
Ankit Mandal, Yash Tiwari, Prasanta K. Panigrahi, Mayukha Pal
Summary: It has been demonstrated that synchronizing physical prior with a neural network reduces training requirements for learning non-linear physical systems. Recent research shows that parameterizing Lagrangian and Hamiltonian using neural network weights and biases, and then executing the equations of motion, leads to more efficient prediction of non-linear dynamical systems compared to conventional neural networks.
CHAOS SOLITONS & FRACTALS
(2022)
Article
Mathematics, Interdisciplinary Applications
Josefine Bohr Brask, Samuel Ellis, Darren P. Croft
Summary: Research on animal social networks involves studying the interactions among individuals in social groups through network analysis. The integration of this field with complex systems could provide valuable insights and solutions to challenges in animal social network research. Collaboration between researchers and biologists studying animal social systems is essential for advancing this interdisciplinary field.
JOURNAL OF COMPLEX NETWORKS
(2021)
Article
Multidisciplinary Sciences
Takahiro Yabe, P. Suresh C. Rao, Satish Ukkusuri, Susan L. Cutter
Summary: With rapid urbanization and increasing climate risks, it is crucial to enhance the resilience of urban systems. However, current studies on disaster resilience often rely on static measures and fail to incorporate the dynamic nature of resilience. This article argues for the use of big data and data-driven complex systems models to quantitatively simulate the recovery trajectories and intrinsic resilience characteristics of communities, paving the way for better policy applications.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)