Guided Exploration for One-way Function Families (GEOFF): Tokenization-Enhanced Reinforcement Learning for Red-Team Password Cracking
Our project focused on how reinforcement learning (RL) could improve password recovery during authorized red team assessments. We trained a model that aims to search (i.e. crack) more efficiently than industry-standard password-cracking tools by iteratively generating meaningful multi-character tokens rather than individual characters.
Interns: Jhanvi Sanwal, Tristan Wang
Mentor: Katie Zaback (AMDS/A4I)
_page-0001.jpg)
-[Autosaved].png)
.png)
.png)
