Inference_for_Graphs_and_Networks.pdf

# Statistically having observed the sample x n 1 x 1 x

• 99

This preview shows pages 37–39. Sign up to view the full content.

Statistically, having observed the sample X n 1 = ( X 1 , . . . , X n ), the prob- lem is to test the hypothesis H k : ν = k 0 that the change has occurred at time k somewhere in the starch of n observations against the alternative H : ν = that there is never a change. The densities of X n 1 given the hypotheses H k and H are p ( X n 1 | H k ) = k i =1 f ( X i | X i 1 1 ) × n i = k +1 g ( X i | X i 1 1 ) , p ( X n 1 | H ) = n i =1 f ( X i | X i 1 1 ) . According to hypothesis testing theory an optimal solution should be based on the likelihood ratio (LR) between these hypotheses given by Λ k n = p ( X n 1 | H k ) p ( X n 1 | H ) = n j = k +1 L j for k < n, L j = g ( X j | X j 1 1 ) f ( X j | X j 1 1 ) , (2.1) where n j = L j = 1 for > n . In the simplest setting, it is assumed that the observations are indepen- dent throughout the entire period of surveillance, so that X 1 , X 2 , . . . , X ν are distributed according to a common pre-change density f , while X ν +1 , X ν +2 , . . . have a common density g f . We refer to this case as the i.i.d. case . Note that due to the i.i.d. assumption, the LR for the n th observation simplifies to L n = g ( X n ) /f ( X n ). The LRs { Λ k n } n 1 are then used by a sequential detection procedure to decide in favor of one of the hypotheses. Given the sequence { X n } n 1 , a sequential detection procedure is defined as a stopping time T adapted to the observations, i.e., the event { T n } is measurable with respect to X 1 , . . . , X n . The stopping time T is nothing but the time instant at which the detection procedure stops and declares that a change is in effect. If the change is declared prematurely, i.e., T ν , then a false alarm is 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.

38 A. G. Tartakovsky (a) Process of interest with a change in the mean. (b) Two possible detection scenarios: false alarm (gray) and correct detection (black). Fig. 2.1. Illustration of sequential changepoint detection. raised. Figure 2.1(a) shows a typical example of change in the mean of the process and its detection. The gray trajectory in Figure 2.1(b) illustrates the false alarm situation when the detection statistic exceeds the detection threshold prior to the change occurring, in which case T can be regarded as the (random) run length to the false alarm. The black trajectory in Figure 2.1(b) illustrates true detection when the detection statistic exceeds the detection threshold past the changepoint. Note that the detection delay characterized by the difference T ν is random.
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