This project studies the performance of Large Language Models for curating datasets of vulnerable functions by classifying vulnerability-fixing commits. It assesses Large Language Models trained on these curated datasets for Vulnerability Detection.
This provides empirical evidence to demonstrate the effectiveness of Large Language Models. Through qualitative analysis, the team showed that Large Language Models can automatically curate accurate, diverse, and large datasets of vulnerable
functions. The work done by the team can be used to streamline Vulnerability Detection research efforts and significantly reduce time spent on laborious labelling tasks.
SUPERVISOR:
Calvin Siak
TEAM MEMBERS:
Isaac Choong Zhu En | Tay Kai Zer | Aldrich Tan Kai Rong | Edison Chan Whye Kit | Sng Kai En Ryan
DIPLOMA:
Diploma in Cybersecurity & Digital Forensics
INDUSTRY PARTNER:
DSO National Laboratories