Learning Machine Learning Part 2: Attacking White Box Models

In the previous post, I went through a very brief overview of some machine learning concepts, talked about the Revoke-Obfuscation project, and detailed my efforts at improving the dataset and models for detecting obfuscated PowerShell scripts. That resulted in three separate tuned models for obfuscated PowerShell script detection: a Logistic Regression model with L2/Ridge regularization, a LightGBM Classifier, and a 4-layer fully connected Neural Network. If you’re not familiar with machine learning, I highly recommend reading the first post …

adversarial-ml analytics & intelligence box information security machine machine learning obfuscation powershell