Главная
Study mode:
on
1
Introduction
2
Outline
3
Motivation
4
Needs
5
Visualization
6
Automated Analysis Research
7
Project Structure
8
Motivation for Work
9
Training Data
10
Auto Document Detection
11
Datasets
12
Stack Overflow
13
Superuser
14
Experiment
15
Model Setup
16
Query Setup
17
Query Demo
18
Variable Success
19
Results
20
Custom Model
21
Bayesian Network
22
Socket
23
Proof Message
24
Inference
25
Bayesian Update
26
Accuracy
27
Precision Recall
28
Speed
29
Impact
30
Adaptability
31
Yarra
32
Malware Demographics
33
Matrix Visualization
34
Output
35
Sample
36
API Calls
Description:
Explore a novel approach to malware capability detection in this Black Hat USA 2013 conference talk. Learn about CrowdSource, an open-source machine learning-based reverse engineering tool that leverages millions of technical documents from the web to identify high-level malware functionality. Discover how this DARPA Cyber Fast Track-funded project aims to provide rapid, automated analysis of malware capabilities, including the ability to detect features like screenshot capture, IRC communication, and webcam operation. Gain insights into the tool's innovative features, such as probabilistic capability detection and traceable output with web document citations. Examine the algorithm behind CrowdSource, its training process using web data, and compelling results demonstrating its effectiveness in reverse engineering active malware variants. Understand the potential impact of this tool on improving visibility into the global malware landscape and accelerating the malware analysis process for security practitioners. Read more

CrowdSource - Crowd Trained Machine Learning Model for Malware Capability Detection

Black Hat
Add to list
0:00 / 0:00