Explore a critical examination of machine learning in cybersecurity through this Black Hat conference talk. Delve into the often-overlooked vulnerabilities of machine learning-based malware defenses from an attacker's perspective. Discover how these popular security measures can be compromised and understand the systemic issues within the network security industry that contribute to these weaknesses. Learn about a proposed solution utilizing innovative data sourcing techniques to address these challenges. Gain insights into the experimental setup, machine learning models, feature space variations, and the importance of local data sources in improving security measures. Analyze the results, validate findings, and understand the commonalities in current machine learning approaches. Conclude with a summary that ties together the key points and implications for the future of cybersecurity.
Defeating Machine Learning - What Your Security Vendor Is Not Telling You
Overview
Syllabus
Introduction
Machine Learning in Security
Experimental Setup
Machine Learning Model
Lock Analogy
Feature Space
Variation
Current Machine Learning
Vendor Data
Local Data Source
Results
Validate
Commonality
Revisiting the Demo
Summary
Taught by
Black Hat
Related Courses
-
Defeating Machine Learning - Systemic Deficiencies for Detecting Malware
-
Building Realistic Machine Learning Systems for Security
-
How is Machine Learning and Data Science Transforming Cyber Security
-
Bot vs. Bot for Evading Machine Learning Malware Detection
-
CrowdSource - Crowd Trained Machine Learning Model for Malware Capability Detection
-
Defending Networks with Incomplete Information - A Machine Learning Approach
