Abstract:We propose a novel high-dimensional linear regression estimator: the Discrete Dantzig Selector, which minimizes the number of nonzero regression coefficients subject to a budget on the maximal absolute correlation between the features and residuals. Motivated by the significant advances in integer optimization over the past 10-15 years, we present a Mixed Integer Linear Optimization (MILO) approach to obtain certifiably optimal global solutions to this nonconvex optimization problem. The current state of algorithmics in integer optimization makes our proposal substantially more computationally attractive than the least squares subset selection framework based on integer quadratic optimization, recently proposed in [8] and the continuous nonconvex quadratic optimization framework of [33]. We propose new discrete first-order methods, which when paired with state-of-the-art MILO solvers, lead to good solutions for the Discrete Dantzig Selector problem for a given computational budget. We illustrate that our integrated approach provides globally optimal solutions in significantly shorter computation times, when compared to off-the-shelf MILO solvers. We demonstrate both theoretically and empirically that in a wide range of regimes the statistical properties of the Discrete Dantzig Selector are superior to those of popular $\ell_{1}$-based approaches. We illustrate that our approach can handle problem instances with p = 10,000 features with certifiable optimality making it a highly scalable combinatorial variable selection approach in sparse linear modeling.

Simultaneous analysis of Lasso and Dantzig selector

Average case analysis of Lasso under ultra-sparse conditions

On Model Selection Consistency of Lasso for High-Dimensional Ising Models

Parallelism, uniqueness, and large-sample asymptotics for the Dantzig selector

Adaptive Lasso in high-dimensional settings

Transductive versions of the LASSO and the Dantzig Selector

High-dimensional stochastic optimization with the generalized Dantzig estimator

The constrained Dantzig selector with enhanced consistency

New Error Analysis for Lasso

The Adaptive Lasso and Its Oracle Properties

The Dantzig selector: Recovery of Signal via $\ell_1-α\ell_2$ Minimization

The Discrete Dantzig Selector: Estimating Sparse Linear Models via Mixed Integer Linear Optimization

The Dantzig Selector: Sparse Signals Recovery via l_p-q Minimization

Discussion: A tale of three cousins: Lasso, L2Boosting and Dantzig

The Linearized Alternating Direction Method of Multipliers for Dantzig Selector

Dual Lasso Selector

Lasso with Latents: Efficient Estimation, Covariate Rescaling, and Computational-Statistical Gaps

Simultaneous Inference for High-Dimensional Linear Models

Two step estimations via the Dantzig selector for models of stochastic processes with high-dimensional parameters

Accuracy guarantees for L1-recovery

On robust width property for Lasso and Dantzig selector