SBIR Phase I: Enabling Robust Binary Code AI via Novel Disassembly

SBIR 第一阶段：通过新颖的反汇编实现鲁棒的二进制代码 AI

• 批准号: 2112109
• 负责人: Xunchao Hu
• 金额: $ 25.57万
• 依托单位: DEEPBITS TECHNOLOGY LLC
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-08-01 至 2022-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2112109&HistoricalAwards=false
• 关键词: SBIR; Phase; Enabling; Robust; Binary

The broader impact of this Small Business Innovation Research (SBIR) Phase I project is to enable novel binary code AI applications for cybersecurity, by providing a fast and accurate method to reverse-engineer executable code. Current solutions are slow and inaccurate, and the volume of code analyzed by cybersecurity applications is huge. Cybersecurity companies either deploy tremendous computing resources to handle huge volumes of code or only extract superficial features from code for AI models, making those solutions extremely vulnerable to adversarial attacks. The proposed solution will not only save computing resources but will allow for improved solutions by extracting complex features from executable code and applying more advanced AI models.This Small Business Innovation Research (SBIR) Phase I project provides an innovative disassembly solution for binary code. The lack of a fast and accurate disassembler has become an obstacle to the applications of novel AI approaches in the cybersecurity industry. The proposed solution combines state-of-the-art binary analysis techniques and newly emerging deep learning techniques to build a fast and accurate disassembler. The proposed solution utilizes GPU technology to accelerate the disassembly process. The final disassembly is expected to be over 100 times faster than state-of-the-art disassemblers, while achieving the same or better accuracy.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

这个小型企业创新研究（SBIR）第一阶段项目的更广泛影响是通过提供一种快速准确的方法来逆向工程可执行代码，从而为网络安全提供新颖的二进制代码AI应用程序。目前的解决方案速度缓慢且不准确，网络安全应用程序分析的代码量巨大。 网络安全公司要么部署巨大的计算资源来处理大量代码，要么只从AI模型的代码中提取表面特征，这使得这些解决方案极易受到对抗性攻击。 该解决方案不仅可以节省计算资源，还可以通过从可执行代码中提取复杂特征并应用更先进的AI模型来改进解决方案。这个小型企业创新研究（SBIR）第一阶段项目提供了一个创新的二进制代码反汇编解决方案。缺乏快速准确的反汇编程序已成为新型人工智能方法在网络安全行业应用的障碍。所提出的解决方案结合了最先进的二进制分析技术和新兴的深度学习技术，以构建快速准确的反汇编程序。 提出的解决方案利用GPU技术来加速拆卸过程。 最终的拆解速度预计将比最先进的拆解机快100倍以上，同时达到相同或更高的精度。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。