Sponsored by
 
Events
News
 
[ Events ]
 
 

Activity Search
Sort out
Field
 
Year
Seminars  
 
NCTS Seminar on Mathematical Biology
 
10:00 - 12:00, January 10, 2020 (Friday)
Lecture Room B, 4th Floor, The 3rd General Building, NTHU
(清華大學綜合三館 4樓B演講室)
Reconstruction of High Dimensional Interaction Networks in Natural Dynamical Systems Based on Empirical Dynamic Modeling
Chun-Wei Chang (NCTS)

Abstract:

Reconstruction of interaction networks from empirical data is challenging, especially when the dimensionality of networks is high, and strength of interactions is not fixed but time-evolving. Therefore, methods quantitatively determining high-dimensional interaction networks for dynamical systems are urgently needed. Here, we reviewed a series of methods based on empirical dynamic modeling (EDM) and proposed one novel approach to quantitatively determine high-dimensional, time-varying interaction networks. We tested our new method using model and empirical time series and demonstrated that interaction strength of reconstructed network is strongly correlated with theoretical expectations (i.e., Jacobian matrices). Although this method is designed to reconstruct interaction networks, it also effectively forecast the dynamical behaviors of network nodes for in both model and empirical data. Our method also reliably infers stability indices of dynamical network, including Lyapunov exponent and structural stability, and preserves various topological properties of weighted network (e.g., centrality and degree distribution) Despite numbers and strength of weak interactions cannot be reliably determined (left-truncated), our novel method based on EDM successfully overcomes the curse of dimensionality in network reconstruction, and thus can be used to evaluate key properties of high-dimensional, time-evolving systems.



 

back to list  
 (C) 2021 National Center for Theoretical Sciences