A Survey of Hardening Neural Code Model

PhD Qualifying Examination


Title: "A Survey of Hardening Neural Code Model"

by

Mr. Zongjie LI


Abstract:

NEURAL code models have been increasingly successful in recent times,
facilitating significant advancements across various code-related applications.
The growing capabilities of these models have led to their integration into
realworld applications, resulting in concerns about their accuracy,
performance, and intellectual property. As a result, there is a growing need to
enhance the robustness and security of neural code models.

This survey is divided into three parts. The first part provides an overview of
neural code models that follow different paradigms. These models come in
different sizes and follow various training and inferencing strategies. In the
second part, we identify the key aspects that require hardening across various
code-related tasks. Finally, in the last part, we highlight the current gap in
the field of reliable and secure neural code models and suggest future
directions for hardening neural code models.


Date:                   Wednesday, 31 January 2024

Time:                   1:00pm - 3:00pm

Venue:                  Room 5501
                        Lifts 25/26

Committee Members:      Dr. Shuai Wang (Supervisor)
                        Dr. Hao Chen (Chairperson)
                        Dr. Dongdong She
                        Dr. Binhang Yuan


**** ALL are Welcome ****