[Colloquium] [Staff] : Reducing LLM Hallucination in Program Analysis Tasks- Tuesday, November 5th

[Abigail Santana via Colloquium]

UNIVERSITY OF CHICAGO
COMPUTER SCIENCE DEPARTMENT
PRESENTS

Xiangyu Zhang
Purdue University
Samuel Conte Professor

Tuesday, November 5th
12:30 pm - 1:30 pm
In Person: John Crerar Library Rm 298

Zoom link: https://uchicago.zoom.us/j/97013580765?pwd=7ATw1tnJYageS2SkhZfYV8KmMD2vVn.1#success
Meeting ID: 970 1358 0765
Passcode: 783088


Title: Reducing LLM Hallucination in Program Analysis Tasks


Abstract: In this talk, I will present our recent efforts in reducing LLM hallucination in program analysis tasks such as

decompilation, data-flow analysis, and bug finding. Although many have started to use LLMs and Code-Language models

in program analysis and program transformation tasks, the results haven't met our expectations. The reason is that these

large models hallucinate a lot in complex tasks. There are various reasons behind this. For example, these models treat

programs no different from natural language texts during pretraining, although the former have a fundamentally different nature (e.g.,

due to loops, recursions, and modular design). In addition, the models usually have limited input sizes, which are insufficient for

complex tasks. I will present a few methods we have developed to reduce hallucination in program analysis, including a novel

pre-taining method that challenges the model to understand program semantics by understanding data-flow, a novel context propagation

method that addresses model input limits, and a new end-to-end LLM based bug detection pipeline that does not directly prompt the

LLM to find bugs, but rather requests the LLM to synthesize code to perform deterministic detection and result sanitization.


Bio: Xiangyu Zhang is a Samuel Conte Professor at Purdue specializing in AI security, software analysis and cyber forensics. His work involves developing techniques to detect bugs, including security vulnerabilities, in traditional software systems as well as AI models and systems, and to leverage AI techniques to perform software engineering and cybersecurity tasks. He has served as the Principal Investigator (PI) for numerous projects funded by organizations such as DARPA, IARPA, ONR, NSF, AirForce, and industry.

[cid:621feec0-742a-416a-a329-528aa938be8a]

Host: Kexin Pei

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20241029/02b75c1f/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Xiangyu Zhang headshot.png
Type: image/png
Size: 87331 bytes
Desc: Xiangyu Zhang headshot.png
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20241029/02b75c1f/attachment-0001.png>