Explore a research presentation from USENIX Security '24 that investigates the effectiveness of Large Language Models (LLMs) in code analysis tasks. Delve into a systematic evaluation conducted by researchers from the University of California, Davis and Temple University examining LLMs' capabilities in processing and analyzing programming code, with particular attention to obfuscated code scenarios. Learn about real-world case studies demonstrating practical applications of LLMs in code analysis, while understanding both their potential benefits and inherent limitations. Gain insights from this 17-minute talk that bridges the gap in existing literature by providing a comprehensive assessment of how LLMs perform in automated code analysis tasks, contributing valuable knowledge to this emerging field of study.
Large Language Models for Code Analysis - Evaluating LLM Performance and Limitations
Overview
Syllabus
USENIX Security '24 - Large Language Models for Code Analysis: Do LLMs Really Do Their Job?
Taught by
USENIX
Related Courses
-
PentestGPT: Evaluating and Harnessing Large Language Models for Automated Penetration Testing
-
Instruction Backdoor Attacks Against Customized Large Language Models
-
Finetuning Large Language Models for Security Log Detections
-
Defeating Secure Code Review GPT Hallucinations - Understanding LLM Limitations in Code Analysis
-
Fuzzing BusyBox - Leveraging LLM and Crash Reuse for Embedded Bug Unearthing
-
Privacy and Security of Large Language Models - Risks and Mitigation
