Eun-Sun Cho

The paper proposes a novel framework to analyze large, obfuscated binaries by decomposing them into structurally coherent units, enabling large-scale dataset generation for LLM-based analysis.

The paper introduces VMPredator, an automated tool that analyzes and deobfuscates virtualization obfuscation in malware by extracting semantic units, successfully restoring program functionality with significant reduction in code length.