Collection of general data mining briefings

1
Data Mining for Malicious Code
Detection and
Security Applications
Prof. Bhavani Thuraisingham The University of
Texas at Dallas
March 17, 2008 Lecture 19

2
Outline

• Acknowledgement
• Prof. Latifur Khan, Dr. Awad (former PhD student
  and postdoc at UTD), Mehedy Masud (PhD student)
• Overview of Data Mining
• Vision for Assured Information Sharing (Our
  framework)
• Data Mining for Cyber security applications
• Intrusion Detection
• Data Mining for Assembly code
• Data mining for Buffer overflow
• Data mining for firewall policy checking
• Data mining for email work detection

3
Vision Assured Information Sharing
Data/Policy for Coalition
Publish
Publish
Data/Policy
Data/Policy
Publish
Data/Policy
Component
Component
Data/Policy for
Data/Policy for
Agency A
Agency C

• Friendly partners
• Semi-honest partners
• Untrustworthy partners

Component
Data/Policy for
Agency B
4
What is Data Mining?
5
Whats going on in data mining?

• What are the technologies for data mining?
• Database management, data warehousing, machine
  learning, statistics, pattern recognition,
  visualization, parallel processing
• What can data mining do for you?
• Data mining outcomes Classification, Clustering,
  Association, Anomaly detection, Prediction,
  Estimation, . . .
• How do you carry out data mining?
• Data mining techniques Decision trees, Neural
  networks, Market-basket analysis, Link analysis,
  Genetic algorithms, . . .
• What is the current status?
• Many commercial products mine relational
  databases
• What are some of the challenges?
• Mining unstructured data, extracting useful
  patterns, web mining, Data mining, security and
  privacy

6
Data Mining for Intrusion Detection Problem

• An intrusion can be defined as any set of
  actions that attempt to compromise the integrity,
  confidentiality, or availability of a resource.
• Attacks are
• Host-based attacks
• Network-based attacks
• Intrusion detection systems are split into two
  groups
• Anomaly detection systems
• Misuse detection systems
• Use audit logs
• Capture all activities in network and hosts.
• But the amount of data is huge!

7
Misuse Detection

• Misuse Detection

8
Problem Anomaly Detection

• Anomaly Detection

9
Our Approach Overview
Training Data
Class
Hierarchical Clustering (DGSOT)
Testing
SVM Class Training
DGSOT Dynamically growing self organizing tree
Testing Data
10
Our Approach Hierarchical Clustering
Our Approach
Hierarchical clustering with SVM flow chart
11
Results
Training Time, FP and FN Rates of Various
Methods
 
12
Introduction Detecting Malicious Executables
using Data Mining

• What are malicious executables?
• Harm computer systems
• Virus, Exploit, Denial of Service (DoS), Flooder,
  Sniffer, Spoofer, Trojan etc.
• Exploits software vulnerability on a victim
• May remotely infect other victims
• Incurs great loss. Example Code Red epidemic
  cost 2.6 Billion
• Malicious code detection Traditional approach
• Signature based
• Requires signatures to be generated by human
  experts
• So, not effective against zero day attacks

13
State of the Art in Automated Detection

• Automated detection approaches
• Behavioural analyse behaviours like source,
  destination address, attachment type, statistical
  anomaly etc.
• Content-based analyse the content of the
  malicious executable
• Autograph (H. Ah-Kim CMU) Based on automated
  signature generation process
• N-gram analysis (Maloof, M.A. et .al.) Based on
  mining features and using machine learning.

14
Our New Ideas (Khan, Masud and Thuraisingham)

• Content -based approaches consider only
  machine-codes (byte-codes).
• Is it possible to consider higher-level source
  codes for malicious code detection?
• Yes Diassemble the binary executable and
  retrieve the assembly program
• Extract important features from the assembly
  program
• Combine with machine-code features

15
Feature Extraction

• Binary n-gram features
• Sequence of n consecutive bytes of binary
  executable
• Assembly n-gram features
• Sequence of n consecutive assembly instructions
• System API call features
• DLL function call information

16
The Hybrid Feature Retrieval Model

• Collect training samples of normal and malicious
  executables.
• Extract features
• Train a Classifier and build a model
• Test the model against test samples

17
Hybrid Feature Retrieval (HFR)

• Training

18
Hybrid Feature Retrieval (HFR)

• Testing

19
Feature Extraction

• Binary n-gram features
• Features are extracted from the byte codes in the
  form of n-grams, where n 2,4,6,8,10 and so on.
  
• Example
• Given a 11-byte sequence 0123456789abcdef012
  345,
• The 2-grams (2-byte sequences) are 0123, 2345,
  4567, 6789, 89ab, abcd, cdef, ef01, 0123, 2345
• The 4-grams (4-byte sequences) are 01234567,
  23456789, 456789ab,...,ef012345 and so on....
• Problem
• Large dataset. Too many features (millions!).
• Solution
• Use secondary memory, efficient data structures
• Apply feature selection

20
Feature Extraction

• Assembly n-gram features
• Features are extracted from the assembly programs
  in the form of n-grams, where n 2,4,6,8,10 and
  so on.
• Example
• three instructions
• push eax mov eax, dword0f34 add ecx,
  eax
• 2-grams
• (1) push eax mov eax, dword0f34
• (2) mov eax, dword0f34 add ecx, eax
• Problem
• Same problem as binary
• Solution
• Same solution

21
Feature Selection

• Select Best K features
• Selection Criteria Information Gain
• Gain of an attribute A on a collection of
  examples S is given by

22
Experiments

• Dataset
• Dataset1 838 Malicious and 597 Benign
  executables
• Dataset2 1082 Malicious and 1370 Benign
  executables
• Collected Malicious code from VX Heavens
  (http//vx.netlux.org)
• Disassembly
• Pedisassem ( http//www.geocities.com/sangcho/ind
  ex.html )
• Training, Testing
• Support Vector Machine (SVM)
• C-Support Vector Classifiers with an RBF kernel

23
Results

• HFS Hybrid Feature Set
• BFS Binary Feature Set
• AFS Assembly Feature Set

24
Results

• HFS Hybrid Feature Set
• BFS Binary Feature Set
• AFS Assembly Feature Set

25
Results

• HFS Hybrid Feature Set
• BFS Binary Feature Set
• AFS Assembly Feature Set

26
Future Plans

• System call
• seems to be very useful.
• Need to Consider Frequency of call
• Call sequence pattern (following program path)
• Actions immediately preceding or after call
• Detect Malicious code by program slicing
• requires analysis

27
Data Mining for Buffer Overflow Introduction

• Goal
• Intrusion detection.
• e.g. worm attack, buffer overflow attack.
• Main Contribution
• 'Worm' code detection by data mining coupled with
  'reverse engineering'.
• Buffer overflow detection by combining data
  mining with static analysis of assembly code.

28
Background

• What is 'buffer overflow'?
• A situation when a fixed sized buffer is
  overflown by a larger sized input.
• How does it happen?
• example

........ char buff100 gets(buff) ........
buff
Stack
memory
Input string
29
Background (cont...)

• Then what?

buff
Stack
........ char buff100 gets(buff) ........
buff
Stack
memory
Return address overwritten
Attacker's code
buff
Stack
memory
New return address points to this memory location
30
Background (cont...)

• So what?
• Program may crash or
• The attacker can execute his arbitrary code
• It can now
• Execute any system function
• Communicate with some host and download some
  'worm' code and install it!
• Open a backdoor to take full control of the
  victim
• How to stop it?

31
Background (cont...)

• Stopping buffer overflow
• Preventive approaches
• Detection approaches
• Preventive approaches
• Finding bugs in source code. Problem can only
  work when source code is available.
• Compiler extension. Same problem.
• OS/HW modification
• Detection approaches
• Capture code running symptoms. Problem may
  require long running time.
• Automatically generating signatures of buffer
  overflow attacks.

32
CodeBlocker (Our approach)

• A detection approach
• Based on the Observation
• Attack messages usually contain code while normal
  messages contain data.
• Main Idea
• Check whether message contains code
• Problem to solve
• Distinguishing code from data

33
Some Statistics

• Statistics to support this observation(a)on
  Windows platforms
• most web servers (port 80) accept data only
• remote access services (ports 111, 137, 138, 139)
  accept data only Microsoft SQL Servers (port
  1434) accept data only
• workstation services (ports 139 and 445) accept
  data only.
• (b) On Linux platforms, most
• Apache web servers (port 80) accept data only
• BIND (port 53) accepts data only
• SNMP (port 161) accepts data only
• most Mail Transport (port 25) accepts data only
• Database servers (Oracle, MySQL, PostgreSQL) at
  ports 1521, 3306 and 5432 accept data only.

34
Severity of the problem

• It is not easy to detect actual instruction
  sequence from a given string of bits

35
Our solution

• Apply data mining.
• Formulate the problem as a classification problem
  (code, data)
• Collect a set of training examples, containing
  both instances
• Train the data with a machine learning algorithm,
  get the model
• Test this model against a new message

36
CodeBlocker Model
37
Feature Extraction
38
Disassembly

• We apply SigFree tool
• implemented by Xinran Wang et al. (PennState)

39
Feature extraction

• Features are extracted using
• N-gram analysis
• Control flow analysis
• N-gram analysis

What is an n-gram? -Sequence of n
instructions Traditional approach -Flow of
control is ignored 2-grams are 02, 24, 46,...,CE
Assembly program
Corresponding IFG
40
Feature extraction (cont...)

• Control-flow Based N-gram analysis

What is an n-gram? -Sequence of n
instructions Proposed Control-flow based
approach -Flow of control is
considered 2-grams are 02, 24, 46,...,CE, E6
Assembly program
Corresponding IFG
41
Feature extraction (cont...)

• Control Flow analysis. Generated features
• Invalid Memory Reference (IMR)
• Undefined Register (UR)
• Invalid Jump Target (IJT)
• Checking IMR
• A memory is referenced using register addressing
  and the register value is undefined
• e.g. mov ax, dx 5
• Checking UR
• Check if the register value is set properly
• Checking IJT
• Check whether jump target does not violate
  instruction boundary

42
Putting it together

• Why n-gram analysis?
• Intuition in general, disassembled executables
  should have a different pattern of instruction
  usage than disassembled data.
• Why control flow analysis?
• Intuition there should be no invalid memory
  references or invalid jump targets.
• Approach
• Compute all possible n-grams
• Select best k of them
• Compute feature vector (binary vector) for each
  training example
• Supply these vectors to the training algorithm

43
Experiments

• Dataset
• Real traces of normal messages
• Real attack messages
• Polymorphic shellcodes
• Training, Testing
• Support Vector Machine (SVM)

44
Results

• CFBn Control-Flow Based n-gram feature
• CFF Control-flow feature

45
Novelty, Advantages, Limitations, Future

• Novelty
• We introduce the notion of control flow based
  n-gram
• We combine control flow analysis with data mining
  to detect code / data
• Significant improvement over other methods (e.g.
  SigFree)
• Advantages
• Fast testing
• Signature free operation
• Low overhead
• Robust against many obfuscations
• Limitations
• Need samples of attack and normal messages.
• May not be able to detect a completely new type
  of attack.
• Future
• Find more features
• Apply dynamic analysis techniques
• Semantic analysis

46
Analysis of Firewall Policy Rules Using Data
Mining Techniques

• Firewall is the de facto core technology of
  todays network security
• First line of defense against external network
  attacks and threats
• Firewall controls or governs network access by
  allowing or denying the incoming or outgoing
  network traffic according to firewall policy
  rules.
• Manual definition of rules often result in in
  anomalies in the policy
• Detecting and resolving these anomalies manually
  is a tedious and an error prone task
• Solutions
• Anomaly detection
• Theoretical Framework for the resolution of
  anomaly
• A new algorithm will simultaneously detect and
  resolve any anomaly that is present in the
  policy rules
• Traffic Mining Mine the traffic and detect
  anomalies

47
Traffic Mining

• To bridge the gap between what is written in the
  firewall policy rules and what is being observed
  in the network is to analyze traffic and log of
  the packets traffic mining
• Network traffic trend may show that some rules
  are out-dated or not used recently

Firewall Policy Rule
48

• Traffic Mining Results

1 TCP,INPUT,129.110.96.117,ANY,...,80,DENY 2
TCP,INPUT,...,ANY,...,80,ACCEPT 3
TCP,INPUT,...,ANY,...,443,DENY 4
TCP,INPUT,129.110.96.117,ANY,...,22,DENY 5
TCP,INPUT,...,ANY,...,22,ACCEPT 6
TCP,OUTPUT,129.110.96.80,ANY,...,22,DENY 7
UDP,OUTPUT,...,ANY,...,53,ACCEPT 8
UDP,INPUT,...,53,...,ANY,ACCEPT 9
UDP,OUTPUT,...,ANY,...,ANY,DENY 10
UDP,INPUT,...,ANY,...,ANY,DENY 11
TCP,INPUT,129.110.96.117,ANY,129.110.96.80,22,DENY
12 TCP,INPUT,129.110.96.117,ANY,129.110.96.80,80
,DENY 13 UDP,INPUT,...,ANY,129.110.96.80,ANY,
DENY 14 UDP,OUTPUT,129.110.96.80,ANY,129.110.10.
,ANY,DENY 15 TCP,INPUT,...,ANY,129.110.96.80,
22,ACCEPT 16 TCP,INPUT,...,ANY,129.110.96.80,
80,ACCEPT 17 UDP,INPUT,129.110..,53,129.110.96.
80,ANY,ACCEPT 18 UDP,OUTPUT,129.110.96.80,ANY,129
.110..,53,ACCEPT
Rule 1, Rule 2 gt GENRERALIZATION Rule 1, Rule
16 gt CORRELATED Rule 2, Rule 12 gt
SHADOWED Rule 4, Rule 5 gt GENRERALIZATION Rule
4, Rule 15 gt CORRELATED Rule 5, Rule 11
gt SHADOWED
Anomaly Discovery Result
49
Worm Detection Introduction

• What are worms?
• Self-replicating program Exploits software
  vulnerability on a victim Remotely infects other
  victims
• Evil worms
• Severe effect Code Red epidemic cost 2.6
  Billion
• Goals of worm detection
• Real-time detection
• Issues
• Substantial Volume of Identical Traffic, Random
  Probing
• Methods for worm detection
• Count number of sources/destinations Count
  number of failed connection attempts
• Worm Types
• Email worms, Instant Messaging worms, Internet
  worms, IRC worms, File-sharing Networks worms
• Automatic signature generation possible
• EarlyBird System (S. Singh -UCSD) Autograph (H.
  Ah-Kim - CMU)

50
Email Worm Detection using Data Mining
Task given some training instances of both
normal and viral emails, induce a hypothesis
to detect viral emails.
We used Naïve Bayes SVM
Outgoing Emails
The Model
Test data
Feature extraction
Classifier
Machine Learning
Training data
Clean or Infected ?
51
Assumptions

• Features are based on outgoing emails.
• Different users have different normal
  behaviour.
• Analysis should be per-user basis.
• Two groups of features
• Per email (of attachments, HTML in body,
  text/binary attachments)
• Per window (mean words in body, variable words in
  subject)
• Total of 24 features identified
• Goal Identify normal and viral emails based
  on these features

52
Feature sets

• Per email features
• Binary valued Features
• Presence of HTML script tags/attributes
  embedded images hyperlinks
• Presence of binary, text attachments MIME types
  of file attachments
• Continuous-valued Features
• Number of attachments Number of words/characters
  in the subject and body
• Per window features
• Number of emails sent Number of unique email
  recipients Number of unique sender addresses
  Average number of words/characters per subject,
  body average word length Variance in number of
  words/characters per subject, body Variance in
  word length
• Ratio of emails with attachments

53
Data Mining Approach
Classifier
Clean/ Infected
Test instance
Clean/ Infected
infected?
SVM
Naïve Bayes
Test instance
Clean?
Clean
54
Data set

• Collected from UC Berkeley.
• Contains instances for both normal and viral
  emails.
• Six worm types
• bagle.f, bubbleboy, mydoom.m,
• mydoom.u, netsky.d, sobig.f
• Originally Six sets of data
• training instances normal (400) five worms
  (5x200)
• testing instances normal (1200) the sixth worm
  (200)
• Problem Not balanced, no cross validation
  reported
• Solution re-arrange the data and apply
  cross-validation

55
Our Implementation and Analysis

• Implementation
• Naïve Bayes Assume Normal distribution of
  numeric and real data smoothing applied
• SVM with the parameter settings one-class SVM
  with the radial basis function using gamma
  0.015 and nu 0.1.
• Analysis
• NB alone performs better than other techniques
• SVM alone also performs better if parameters are
  set correctly
• mydoom.m and VBS.Bubbleboy data set are not
  sufficient (very low detection accuracy in all
  classifiers)
• The feature-based approach seems to be useful
  only when we have
• identified the relevant features
• gathered enough training data
• Implement classifiers with best parameter
  settings