Just another copy and paste? Comparing the security vulnerabilities of ChatGPT generated code and StackOverflow answers

no code implementations • 22 Mar 2024 • Sivana Hamer, Marcelo d'Amorim, Laurie Williams

Our findings suggest developers are under-educated on insecure code propagation from both platforms, as we found 274 unique vulnerabilities and 25 types of CWE.

Code Generation Tools (Almost) for Free? A Study of Few-Shot, Pre-Trained Language Models on Code

no code implementations • 2 Jun 2022 • Patrick Bareiß, Beatriz Souza, Marcelo d'Amorim, Michael Pradel

For example, we find that providing a small natural language description of the code generation task is an easy way to improve predictions.

Code Generation Few-Shot Learning +1

Shipwright: A Human-in-the-Loop System for Dockerfile Repair

1 code implementation • 3 Mar 2021 • Jordan Henkel, Denini Silva, Leopoldo Teixeira, Marcelo d'Amorim, Thomas Reps

Furthermore, in a "time-travel" analysis of broken Dockerfiles that were later fixed, we found that SHIPWRIGHT proposed repairs that were equivalent to human-authored patches in 22. 77% of the cases we studied.

Language Modelling Software Engineering

Exposing Bugs in JavaScript Engines through Test Transplantation and Differential Testing

1 code implementation • 7 Dec 2020 • Igor Lima, Jefferson Silva, Breno Miranda, Gustavo Pinto, Marcelo d'Amorim

This paper investigates the use of test transplantation and differential testing for revealing functional bugs in JavaScript engines.

Software Engineering

Generating Adversarial Inputs Using A Black-box Differential Technique

no code implementations • 10 Jul 2020 • João Batista Pereira Matos Juúnior, Lucas Carvalho Cordeiro, Marcelo d'Amorim, Xiaowei Huang

Algorithmically, DAEGEN uses a local search-based optimization algorithm to find DIAEs by iteratively perturbing an input to maximize the difference of two models on predicting the input.

Adversarial Attack