27-09-2016, 02:17 PM
DETECTING ANDROID MALWARE WITH FEATURE EXTRACTED SOURCE CODE USING PROBABLISTIC DISCRIMINATIVE MODEL
[attachment=71784]
Abstract
Mobile device is an important part of our day to day life. The Android platform has now become a market leader. There exist an number of approaches for detecting malware in the Android devices such as permissions, source code analysis or dynamic analysis.
The system uses Android Application Programming Interface (API) calls for accessing source code from the decompiled source code.
This system uses a probabilistic discriminative model based on the regularized logistic regression for the detection of Android malware’s in Applications
Project objective
The main objective is to develop system that can detect malwares in.apk file
Source Code Feature Extraction is a process of decompilation of .apk files into java source code files.
Probabilistic discriminative model uses training set to classify the positive and negative samples.
Using our system the malware in the application codes can be detected and revocation is done.
Problem definition
The growing popularity of android has made mobile platforms a prime target for attack.
it is worth noting that the arm race between malicious application and malicious application detection technique.
Static analysis approaches focus on comparing programs to known malware based on the program code.
Several approaches are used to detect application is malicious by previous researchers such as using two or three permissions triggers a warning that the application is risky.
Need for proposed methodology
Android has Google Play as the primary store, Kindle uses the Amazon Appstore for Android, iOS has iTunes App Store.This paradigm presents both challenges and opportunities for malware defense.
Applications are generally distributed in a form that can be decompiled into source code .
Decompiled source code of android applications can provide more detailed information than the list of permissions that the application request.
Decompile
Android applications are packed as .apk files. To extract features from them, we first decompile the .apk files into Java source code files, and then extract features from the source code.
Feature granularities
To extract API usage information from the source code as features. One can view the source code as documents, and apply document classification techniques to the problem.
In document classification, term frequency (TF) is typically used as features.
Input parameter: java source code
Output parameter: Application Programming Interface(API) usage informations.
[attachment=71784]
Abstract
Mobile device is an important part of our day to day life. The Android platform has now become a market leader. There exist an number of approaches for detecting malware in the Android devices such as permissions, source code analysis or dynamic analysis.
The system uses Android Application Programming Interface (API) calls for accessing source code from the decompiled source code.
This system uses a probabilistic discriminative model based on the regularized logistic regression for the detection of Android malware’s in Applications
Project objective
The main objective is to develop system that can detect malwares in.apk file
Source Code Feature Extraction is a process of decompilation of .apk files into java source code files.
Probabilistic discriminative model uses training set to classify the positive and negative samples.
Using our system the malware in the application codes can be detected and revocation is done.
Problem definition
The growing popularity of android has made mobile platforms a prime target for attack.
it is worth noting that the arm race between malicious application and malicious application detection technique.
Static analysis approaches focus on comparing programs to known malware based on the program code.
Several approaches are used to detect application is malicious by previous researchers such as using two or three permissions triggers a warning that the application is risky.
Need for proposed methodology
Android has Google Play as the primary store, Kindle uses the Amazon Appstore for Android, iOS has iTunes App Store.This paradigm presents both challenges and opportunities for malware defense.
Applications are generally distributed in a form that can be decompiled into source code .
Decompiled source code of android applications can provide more detailed information than the list of permissions that the application request.
Decompile
Android applications are packed as .apk files. To extract features from them, we first decompile the .apk files into Java source code files, and then extract features from the source code.
Feature granularities
To extract API usage information from the source code as features. One can view the source code as documents, and apply document classification techniques to the problem.
In document classification, term frequency (TF) is typically used as features.
Input parameter: java source code
Output parameter: Application Programming Interface(API) usage informations.