Explore a groundbreaking machine learning classifier for data-driven vulnerability prioritization in this Black Hat conference talk. Delve into the complexities of modern technology stacks, increasing vulnerability volumes, and the challenges of effective remediation strategies. Learn about measuring coverage and efficiency in vulnerability management, current attacker velocities, and the case for predictive approaches. Gain insights into machine learning applications for cybersecurity, including data sources, algorithm types, and supervised classification. Discover how probability can be leveraged to improve decision-making in vulnerability and risk management, and witness live predictions and fulfillments during this 50-minute session presented by Kenna researchers Michael Roytman and Jonathan Cran.
Those Don't Matter! Effective Prioritization Through Exploit Prediction
Overview
Syllabus
Intro
The Modern Stack is COMPLEX
Vulnerability Volume Increasing
Remember the Recall
What Matters for Scoring
Measuring Remediation Strategies
Coverage & Efficiency, Explained
Coverage / Efficiency Tradeoff
Current Attacker Velocity
Factoring in Velocity
The Case for Prediction
What Is Machine Learning?
Data Sources: CVE Enrichment Projects
Data Sources: Exploit Code & Observations
Type of Algorithms
Supervised Classification
Predictive - The Expectations
Coverage Efficiency Tradeoffs
Machine Learning Has Side Benefits
Lesson: Probability is our friend
Taught by
Black Hat
