Deep Adversarial Architectures for Detecting and Generating Maliciousness

Explore deep adversarial architectures for detecting and generating malicious content in this 39-minute conference talk from BSidesLV 2016. Dive into the world of cybersecurity as Hyrum Anderson presents an overview of deep learning techniques applied to malware detection and generation. Learn about logistic regression, deep learning packages, and the key vulnerabilities in deep learning models. Discover the concepts of red team modeling, autoencoders, and the comparison between shallow and deep learning approaches. Gain insights into hardening techniques, false positives, and vulnerability assessment. Engage with topics such as Deep DJ, Deep TJ, and Deep DGA, and understand their implications for real-world cybersecurity applications.

Intro
Overview
Motivation
Validation
Red Team Model
Outline
Shallow Learning
Logistic Regression
Deep Learning
Deep Learning Packages
Deep Learning Model
Key to Deep Learning
Deep Learning Vulnerability
Application Review
Red vs Blue
Autoencoder
Results
Comparison
Deep DJ
Hardening
Conclusion
Questions
Autoencoders
Deep TJ
Deep DGA
Real Deep Learning
What would I be
False positives
Punic
Vulnerability Assessment