This tool assesses the reproducibility of AI Trojan detector results across different computing environments.
Researchers and developers working on AI security can use this tool to evaluate how Trojan detection algorithms perform across various hardware and software configurations. It helps in understanding the variability in results due to environmental factors, thereby aiding in the development of more robust detection methods.
The original version control history is not available, as this is a ported version of the project from a NIST GitLab server.
Clone the repository using 'git clone https://github.com/johannes-losert/NIST-SURF-TrojAI-2022.git'
Navigate to the 'documentation' folder for detailed instructions
Refer to the 'testing' folder for the software framework used