Analysis and Actions on Graph Data.
dc.contributor.author | Chen, Pin-Yu | |
dc.date.accessioned | 2017-01-26T22:17:45Z | |
dc.date.available | NO_RESTRICTION | |
dc.date.available | 2017-01-26T22:17:45Z | |
dc.date.issued | 2016 | |
dc.date.submitted | ||
dc.identifier.uri | https://hdl.handle.net/2027.42/135752 | |
dc.description.abstract | Graphs are commonly used for representing relations between entities and handling data processing in various research fields, especially in social, cyber and physical networks. Many data mining and inference tasks can be interpreted as certain actions on the associated graphs, including graph spectral decompositions, and insertions and removals of nodes or edges. For instance, the task of graph clustering is to group similar nodes on a graph, and it can be solved by graph spectral decompositions. The task of cyber attack is to find effective node or edge removals that lead to maximal disruption in network connectivity. In this dissertation, we focus on the following topics in graph data analytics: (1) Fundamental limits of spectral algorithms for graph clustering in single-layer and multilayer graphs. (2) Efficient algorithms for actions on graphs, including graph spectral decompositions and insertions and removals of nodes or edges. (3) Applications to deep community detection, event propagation in online social networks, and topological network resilience for cyber security. For (1), we established fundamental principles governing the performance of graph clustering for both spectral clustering and spectral modularity methods, which play an important role in unsupervised learning and data science. The framework is then extended to multilayer graphs entailing heterogeneous connectivity information. For (2), we developed efficient algorithms for large-scale graph data analytics with theoretical guarantees, and proposed theory-driven methods for automatic model order selection in graph clustering. For (3), we proposed a disruptive method for discovering deep communities in graphs, developed a novel method for analyzing event propagation on Twitter, and devised effective graph-theoretic approaches against explicit and lateral attacks in cyber systems. | |
dc.language.iso | en_US | |
dc.subject | graph data analytics for data science and cyber security | |
dc.subject | phase transition analysis of community detection and graph clustering | |
dc.subject | single-layer and multi-layer graphs | |
dc.subject | graph spectral decomposition and eigenvalue decomposition | |
dc.subject | node and edge additions and removals | |
dc.subject | complex network analysis and network science | |
dc.title | Analysis and Actions on Graph Data. | |
dc.type | Thesis | en_US |
dc.description.thesisdegreename | PhD | en_US |
dc.description.thesisdegreediscipline | Electrical & Computer Eng PhD | |
dc.description.thesisdegreegrantor | University of Michigan, Horace H. Rackham School of Graduate Studies | |
dc.contributor.committeemember | Hero III, Alfred O | |
dc.contributor.committeemember | Koutra, Danai | |
dc.contributor.committeemember | Romero, Daniel M | |
dc.contributor.committeemember | Subramanian, Vijay Gautam | |
dc.subject.hlbsecondlevel | Computer Science | |
dc.subject.hlbsecondlevel | Electrical Engineering | |
dc.subject.hlbtoplevel | Engineering | |
dc.description.bitstreamurl | http://deepblue.lib.umich.edu/bitstream/2027.42/135752/1/pinyu_1.pdf | |
dc.identifier.orcid | 0000-0003-1039-8369 | |
dc.identifier.name-orcid | Chen, Pin-Yu; 0000-0003-1039-8369 | en_US |
dc.owningcollname | Dissertations and Theses (Ph.D. and Master's) |
Files in this item
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.