Welcome to Trojai_RL’s documentation!

_images/TrojAI_logo.png _images/apl2.png

trojai_rl is a sister Python module to trojai, designed for quick generation of triggered deep reinforcement learning (DRL) models. It contains two submodules: trojai_rl.datagen contains the triggered learning environments for model training as well as the API for using them with trojai_rl.modelgen. The trojai_rl.modelgen module contains the necessary API functions to generate DRL models from the triggered environments.

Note that this repository is in early prototype stages, and is subject to potential errors and major updates.

Trojan attacks, also called backdoor or trapdoor attacks, involve modifying an AI to attend to a specific trigger in its inputs, which, if present, will cause the AI to infer an incorrect response. For more information, read the Introduction, the documentation for trojai, and our article on arXiv.