This $338,993 Project Grant from the National Science Foundation's Computer and Information Science and Engineering program (CFDA 47.070) supports research at George Mason University to advance semantics-oriented binary code analysis. The University will develop novel deep learning-based approaches to analyze closed-source software at both the instruction and control flow graph levels, with the goal of achieving high accuracy and scalability in binary code understanding. Specifically, the research aims to represent instructions and basic blocks as embeddings to facilitate analysis, leverage captured code semantics at lower levels to analyze control flow graphs, and investigate how the techniques can handle code obfuscations. In addition to propelling applications in vulnerability discovery and malware analysis, the award is intended to foster new research opportunities and involve students from Benedict College in computer security research through outreach activities.