Towards LLM-Based Analysis of Virtualization-Obfuscated Code through Automated Data Generation
密码学与安全
2026-05-12 v1
摘要
Virtualization-based obfuscation produces extremely large and structurally complex binaries, posing challenges for LLM-based analysis due to input size limits and the need for large-scale labeled data. We address this by focusing on structural rather than full semantic analysis. Obfuscated binaries are decomposed into the largest semantically coherent units that fit within LLM constraints and are labeled according to their structural roles. We implement a static analysis framework to automate labeling and enable large-scale dataset generation. Our prototype shows strong performance on real-world virtualization obfuscators.
引用
@article{arxiv.2605.09961,
title = {Towards LLM-Based Analysis of Virtualization-Obfuscated Code through Automated Data Generation},
author = {Sangjun An and Hyeyeon Park and Yejin Son and Seoksu Lee and Eun-Sun Cho},
journal= {arXiv preprint arXiv:2605.09961},
year = {2026}
}