Inference_for_Graphs_and_Networks.pdf

Against alternatives that can be formed by

• 99

This preview shows pages 79–81. Sign up to view the full content.

) against alternatives that can be formed by restricting the overall parameter space, Θ , to a subset Θ A Θ . The generalized likelihood ratio test statistic (GLRT) is a natural statistic to use. Let λ γ = 2 log L ( θ 0 ( γ ) | x ( γ )) sup θ Θ A L ( θ ( γ ) | x ( γ )) . (3.1) The size of λ γ depends on the number of parameters being tested in the window, making it diﬃcult to use directly. To address this issue, we nor- malize λ γ by converting it into a p -value, p γ . These p -values are discussed in detail in Section 3.4.5. To scan for anomalies in the T ime × Graph product space, we must slide over all windows γ , keeping track of the scan statistic Ψ = min γ p γ . In practice, thresholding of the set of p -values is performed so more than just the minimum p -value can be considered. For online monitoring, we set a threshold on the p -values to control the false discovery rate (Benjamini and Hochberg, 1995). This threshold setting is described in more detail in Section 3.4.6, but we emphasize that generally, when a detection occurs, a set of windows (not just one) are detected, and so the union of these windows is the detected anomaly produced by the system. 3.3. Independence Among Edges in a Path In order to scan for anomalous shapes, it is necessary to have models that describe the behavior of the data in the window γ under normal conditions. The number of enumerated subgraphs tends to scale exponentially with the number of nodes, however, and an assumption of independence among the edges in the shape facilitates scaling the computations required to process large graphs at line speeds, under reasonable memory requirements. While the general approach discussed in Section 3.2.1 does not require indepen- dence among the edges in the window γ , independence will be assumed among the stars and paths discussed in Section 3.1.3 and used in the simulation and real-data sections. Note that for those shapes, no edge is repeated. Copyright © 2014. Imperial College Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law. EBSCO Publishing : eBook Collection (EBSCOhost) - printed on 2/16/2016 3:37 AM via CGC-GROUP OF COLLEGES (GHARUAN) AN: 779681 ; Heard, Nicholas, Adams, Niall M..; Data Analysis for Network Cyber-security Account: ns224671

This preview has intentionally blurred sections. Sign up to view the full version.

80 J. Neil, C. Storlie, C. Hash and A. Brugh Edge independence ensures the ability to scale, since models (and the storage of their parameters) for each edge are suﬃcient to construct models for subgraphs of edges under this assumption, whereas non-independence might require models for each shape , of which there may be many hundreds of millions, if not billions. Therefore, an examination of the assumption of independence among edges connected in a path is conducted.
This is the end of the preview. Sign up to access the rest of the document.
• Spring '12
• Kushal Kanwar
• Graph Theory, Statistical hypothesis testing, Imperial College Press, applicable copyright law

{[ snackBarMessage ]}

What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern