The Role of Hyperparameters in Machine Learning Models and How to Tune Them
Demonstrating why hyperparameter tuning and documentation should be standard in ML robustness checks.
Machine Learning
Private Post-GAN Boosting
Differentially private method combining samples from GAN training to create high-quality synthetic data.
Really Useful Synthetic Data -- A Framework to Evaluate the Quality of Differentially Private Synthetic Data
A framework to evaluate the quality of differentially private synthetic data from an applied researcher’s perspective.
How Cross-Validation Can Go Wrong and What to Do About it
Demonstrating the severe consequences of problematic cross-validation usage in predictive modeling.
Deep Multiple Imputation - Using Mixture Density Networks to Impute Missing Values
Poster presentation at PolMeth 2018 on using deep learning for missing data imputation.