Feature Selection from Differentially Private Correlations

Ryan Swope, Amol Khanna, Philip Doldo +2

Data scientists often seek to identify the most important features in high-dimensional datasets. This can be done through $L_1$-regularized regression, but this can become ineffici…