Modeling and Analysis of Intentional And Unintentional Security Vulnerabilities in a Mobile Platform Page: 48
xi, 149 pages : illustrations (chiefly color)View a full description of this dissertation.
Extracted Text
The following text was automatically extracted from the image on this page using optical character recognition software:
Clustering algorithm used- Sensitivity Specificity Balanced Accuracy
in intention identification Accuracy
k-mean 77% 68% 72.5% 70%
EM 50% 100% 75% 89%
TABLE 6.2. Overall potential malware identification performance when unsuper-
vised learning (clustering) is utilized. Due to better identification of benign apps in
EM, it produced a higher accuracy than k-mean as the dataset had more malware
samples than benign samples.
with lower accuracies. GoldDream malware family is the only family in the database that belongs
to both repackaging and stand-alone. My algorithm identified all the samples in this family as well.
In overall, this algorithm identified 50% of the repackaged malware apps and 55% of standalone
malware apps.
6.3. Related Work
Rassameeroj and Tanahashi[84] clustered apps based on their permission requests. It was
concluded that it is possible to detect malicious apps based on permission requests as long as there
is a careful selection of the permission set. D. Barrera et al. explains how Android permission
can be utilized in mobile security [9]. They utilized Self-Organizing Map (SOM) to evaluate the
access control permission requests. Their analysis was based on about 1,100 Android apps to study
the permission usage patterns and they claimed that certain permissions are very frequently used,
while some are not.
A. Shabtai et al. also adopted a machine learning model to classify android apps into two
groups (tools and games) with a dataset of 2,850 apps [91]. R. Perdisci and M. U's showed, how
unsupervised learning can be used to perform malware clustering and how it can be evaluated. The
database consisted of about 3,000 malware samples.[75].
Apps in Facebook also follow a permission based access control. M. Frank et al. utilized a
probabilistic model to mine permission request patterns from Android and Facebook applications
with a large number of samples [41]. Though their goal was not to identify malware samples, they
used an unsupervised learning model to find permission request patterns. One of their findings
indicated that Android app categories are related to its permission request patterns. This fortifies
my own argument that similar task-intention apps are related to its permission request patterns.48
Upcoming Pages
Here’s what’s next.
Search Inside
This dissertation can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Dissertation.
Mohamed Issadeen, Mohamed Fazeen. Modeling and Analysis of Intentional And Unintentional Security Vulnerabilities in a Mobile Platform, dissertation, December 2014; Denton, Texas. (https://digital.library.unt.edu/ark:/67531/metadc700067/m1/62/: accessed July 18, 2024), University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu; .