M.S. (Master of Science)
Department of Mechanical Engineering
Network representation provides a natural framework for the study of real world complex systems. From social networks and animal groups to interneuronal communications and power grid systems, complex patterns of interaction can be captured and modeled using networks in a simple mathematical form. In many cases, however, a faithful network representation of the system is not readily available. For this reason, network reconstruction has become a growing topic of interest in recent years, the goal of which is to discover the hidden interaction patterns among individuals by ﬁtting input-output data from multiple experiments to candidate network topologies.
During a cascade, however, only one set of data is available to describe the response of each individual; this is common for many real world scenarios when only a single instance of an event can be observed (such as crowd panic or disease epidemics). To reconstruct the< underlying network with limited data, in this thesis we formulated a model-based reconstruction framework which assumes ﬁrst-order dynamics of individual response and linear interactions between individuals. We tested this framework by simulating cascades and performing reconstruction for a number of sample networks. Next, we analyzed the dependence of the number of candidate network solutions on prior knowledge about the network captured in terms of dynamic parameters and connectivity. Our results indicated that the number of valid solutions can be greatly reduced provided some knowledge of nodal parameters and network topology is available. Finally, we validated the modeling framework on experimental data from literature describing ﬁsh response to a simulated predator attack. Speciﬁcally, we conﬁrm that individual ﬁsh startle response follows ﬁrst order dynamics and that the reconstructed topology in a school of escaping ﬁsh approximates their visual perceptual ﬁeld.
Chwistek, Katherine Irena, "Model-Based Network Reconstruction from Cascade Dynamics" (2020). Graduate Research Theses & Dissertations. 6925.
Northern Illinois University
Rights Statement 2
NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.