ICSE 2024 (series) / Artifact Evaluation /
Dataflow Analysis-Inspired Deep Learning for Efficient Vulnerability Detection
Title: Dataflow Analysis-Inspired Deep Learning for Efficient Vulnerability Detection
Purpose
Train and evaluate DeepDFA, the model proposed in our paper, on the dataset used in our paper
Badges
Available and Reusable
Skills assumed by the reviewer
- Basic Python environment management using e.g. conda
- Basic Linux file system manipulation such as creating/deleting files and directories
- Able to run Bash scripts and supply appropriate parameters
- Able to modify configuration files in formats such as YAML
Environment
NVIDIA GPU capable of training small graph neural networks or CodeBERT-sized transformer models - we used NVIDIA GTX 3090 Linux, Python >3.10