Skip to content

mnhng/GPA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GPA - Generalized Prevalence Adjustment

Healthcare data often come from multiple sites in which the correlations between the target (Y) and the confounding variables (Z) can vary widely. If deep learning models exploit these unstable correlations, they might fail catastrophically in unseen sites. GPA is a flexible method that adjusts model predictions to the shifting correlations between prediction target and confounders to safely exploit unstable features.

GPA learns a stable adaptive predictor by modeling the data-generation process using two separate estimators: one for the stable mechanism and the other for the shifting P(Y|Z) distribution. GPA can infer the interaction between the target and the confounders at new sites using only unlabeled samples from those sites. Furthermore, GPA can better predict for samples without Z at test time even for high-dimensional Z.

Link to paper: Adapting to Shifting Correlations with Unlabeled Data Calibration

How to Setup

  1. Install miniconda
  2. Setup the conda environment
    conda env create -n gpa
    conda activate gpa
    conda install torchvision pytorch-cuda -c pytorch -c nvidia
    conda install numpy scipy scikit-learn pandas

How to Use

  1. Activate the conda environment
  2. Execute the shell scripts in the examples folder.
  3. For ISIC experiment, see the instruction in the data/ISIC folder.
  4. For chest X-Ray experiment, download the datasets from PhysioNet.

Cite

@inproceedings{nguyen2024adapting,
    title={{Adapting to Shifting Correlations with Unlabeled Data Calibration}},
    author={Nguyen, Minh and Wang, Alan Q. and Kim, Heejong and Sabuncu, Mert R.},
    booktitle={Proceedings of ECCV},
    year={2024}
}
@inproceedings{nguyen2024robust,
    title={{Robust Learning via Conditional Prevalence Adjustment}},
    author={Nguyen, Minh and Wang, Alan Q., and Kim, Heejong and Sabuncu, Mert R.},
    booktitle={Proceedings of WACV},
    year={2024}
}

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors