Tasks

Each task defines a problem — its input, public metadata, output, and evaluation metric — that DP algorithms solve. Pick a task to see its formulation and the algorithms benchmarked on it.

• Linear Regression — Estimate regression coefficients β from (X, y) under differential privacy.
• M-Estimation — Generalized empirical-risk minimization — the parent of mean / median / linear regression.
• Mean Estimation — Estimate the mean of bounded data.
• Median Estimation — Estimate the median (a robust location estimate).
• Logistic Regression — Binary classification via the logistic loss.
• Single Parameter Estimation — Estimate one headline regression coefficient — the applied study's reported effect.
• Simple Linear Regression — Fit a line (slope, intercept) to a single scalar covariate — the smallest regression task.