On Semi-Supervised Estimation of Distributions

We study the problem of estimating the joint probability mass function (pmf) over two random variables. In particular, the estimation is based on the observation of $m$ samples containing both variables and $n$ samples missing one fixed variable. We adopt the minimax framework with $l_p^p$ loss functions, and we show that the composition of uni-variate minimax estimators achieves minimax risk with the optimal first-order constant for $p\ge 2$, in the regime $m=o(n)$.