AAFREEN-ZAHRA-KAZMI01/LEAD-credit-attack-detection ? reverse-engineered prompt
Reverse engineered prompt
Build me this credit attack detection project from the paper and make it easy to run end to end.
I want a notebook friendly repo that can reproduce the main experiments for LEAD, the lightweight detector for feature manipulation attacks on tree based credit scoring models. It should load the public credit datasets, train the credit models, run the attack simulations, score applications with the LEAD checks, and generate the tables and figures from the results folder. Please include the quick mode for a faster demo, and the full mode for the paper style run.
Make sure the code can also show the comparison methods, the ablation study, the fairness audit, and the speed test. If anything is missing, use the README and existing files to wire it together as cleanly as possible, and look up current docs online if you need to. I’d like a simple setup path, clear notebook entry points, and scripts that run the experiments from the command line too.
Are you gonna build this?
make sure you review the code using coderabbit