L1 Regularization Paths in Linear Models by Parametric Gaussian Message Passing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Yun-Peng, Loeliger, Hans-Andrea
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910143446253568
author Li, Yun-Peng
Loeliger, Hans-Andrea
author_facet Li, Yun-Peng
Loeliger, Hans-Andrea
contents The paper considers the computation of L1 regularization paths in a state space setting, which includes L1 regularized Kalman smoothing, linear SVM, LASSO, and more. The paper proposes two new algorithms, which are duals of each other; the first algorithm applies to L1 regularization of independent variables while the second applies to L1 regularization of dependent variables. The heart of the proposed algorithms is parametric Gaussian message passing (i.e., Kalman-type forward-backward recursions) in the pertinent factor graphs. The proposed methods are broadly applicable, they (usually) require only matrix multiplications, and their complexity can be competitive with prior methods in some cases.
format Preprint
id arxiv_https___arxiv_org_abs_2604_16949
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle L1 Regularization Paths in Linear Models by Parametric Gaussian Message Passing
Li, Yun-Peng
Loeliger, Hans-Andrea
Machine Learning
Signal Processing
Methodology
The paper considers the computation of L1 regularization paths in a state space setting, which includes L1 regularized Kalman smoothing, linear SVM, LASSO, and more. The paper proposes two new algorithms, which are duals of each other; the first algorithm applies to L1 regularization of independent variables while the second applies to L1 regularization of dependent variables. The heart of the proposed algorithms is parametric Gaussian message passing (i.e., Kalman-type forward-backward recursions) in the pertinent factor graphs. The proposed methods are broadly applicable, they (usually) require only matrix multiplications, and their complexity can be competitive with prior methods in some cases.
title L1 Regularization Paths in Linear Models by Parametric Gaussian Message Passing
topic Machine Learning
Signal Processing
Methodology
url https://arxiv.org/abs/2604.16949