Explore a novel approach for predicting debug information in stripped binaries through this 27-minute conference talk. Learn about using machine learning to train probabilistic models on non-stripped binaries and applying them to predict properties of meaningful elements in unseen stripped binaries. Delve into topics such as recovering variables, predicting names and types, and evaluating prediction accuracy. Gain insights into the challenges of working with stripped binaries and discover how this technique can be applied to malware inspection.
Debin: Predicting Debug Information in Stripped Binaries
Association for Computing Machinery (ACM) via YouTube
Overview
Syllabus
Intro
Binaries with debug symbols
Stripped binaries
Challenges
DeBIN: Recovering debug information
DeBIN: System overview
Learning how to recover variables
Probabilistic graphical model
Learning how to predict names and types
DeBIN implementation
Variable recovery accuracy
Name and type prediction accuracy
Evaluation of name and type prediction
Malware inspection
Summary
Taught by
Association for Computing Machinery (ACM)
