English

Efficient Path Algorithms for Clustered Lasso and OSCAR

Machine Learning 2020-06-17 v1 Machine Learning Computation

Abstract

In high dimensional regression, feature clustering by their effects on outcomes is often as important as feature selection. For that purpose, clustered Lasso and octagonal shrinkage and clustering algorithm for regression (OSCAR) are used to make feature groups automatically by pairwise L1L_1 norm and pairwise LL_\infty norm, respectively. This paper proposes efficient path algorithms for clustered Lasso and OSCAR to construct solution paths with respect to their regularization parameters. Despite too many terms in exhaustive pairwise regularization, their computational costs are reduced by using symmetry of those terms. Simple equivalent conditions to check subgradient equations in each feature group are derived by some graph theories. The proposed algorithms are shown to be more efficient than existing algorithms in numerical experiments.

Keywords

Cite

@article{arxiv.2006.08965,
  title  = {Efficient Path Algorithms for Clustered Lasso and OSCAR},
  author = {Atsumori Takahashi and Shunichi Nomura},
  journal= {arXiv preprint arXiv:2006.08965},
  year   = {2020}
}
R2 v1 2026-06-23T16:21:47.273Z