1: \begin{abstract}

2: This paper investigates tradeoffs among optimization errors, statistical rates of convergence and the effect of heavy-tailed  errors for high-dimensional robust regression with nonconvex regularization. When the additive errors in linear models have only bounded second moment,  we show that iteratively reweighted $\ell_1$-penalized adaptive Huber regression estimator satisfies exponential deviation bounds and oracle properties,  including the oracle convergence rate and variable selection consistency, under a weak beta-min condition. 

3: Computationally, we need as many as $\cO( \log s + \log\log d )$ iterations to reach such an oracle estimator, where $s$ and $d$ denote the sparsity and ambient dimension, respectively. Extension to a general class of robust loss functions is also considered.

4: Numerical studies lend strong support to our methodology and theory. 

5: \end{abstract}

6: