A Machine Learning approach for detecting a Malware
summary
Static static-analysis pipeline extracts VBA macro features (objects, goto counts, library calls, string ops, modules, forms), combines automated and analyst-tagged attributes, and applies classifiers after removing categorical/composite features; dropping harmful features yields an 80% performance gain and lower false positives. A production flow uses a hash prefilter to bypass known samples and sends unknowns to the model. The system catches and blocks real spam runs (example: Aug 21–27 campaign) and reduces reliance on sandbox execution.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:02Machine Learning at Symantec
- 3:04What is Malware
- 8:34Email Ways
- 13:55Handwritten Text
- 15:45Machine Learning
- 17:24Feature Engineering
- 20:37Recent Findings
related talks
51mAdvanced VBA Macros Attack & Defence
30mData Driven Security
48mCatch me if you can: Building a Web Malware Analyzer using Machine Learning
22mMalware Hunting - Using python as an attack weapon
41mHow AI Is Changing the Malware Landscape
45mApplying Machine Learning Methods to Static Analysis of Executables
18mMac-A-Mal: An Automated Platform for Mac Malware Hunting
32mAI/Machine Learning: What is Actually Working in Cybersecurity?