AI Paper Atlas

Research map / Learning theory and algorithms / Matrix and tensor methods

New Insights for the Multivariate Square-Root Lasso

Aaron J. Molstad

JMLR 2022 · Learning theory and algorithms › Matrix and tensor methods · approach: High-dimensional estimation theory

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use cross-validation instead of theory-based quantile tuning of lambda (ablation)MSR · Model 1-3, L1 penalty, M1 betasquared Frobenius norm error: directly tuned estimators tend to perform substantially worse than MSR-CV, including MSR-OrImproved
add refitting (SUR-type likelihood re-estimation) to theory-tuned MSR (ablation)MSR-q95/q85/q75/q50 · Model 1-3, L1 penalty, M1 betarelative squared Frobenius norm error to MSR-CV: performance relative to MSR-CV much improved with refittingImproved
use theory-based tuning (q50) with group lasso penalty (ablation)MSR-q50 · Model 1-3, group lasso, M2 betarelative squared Frobenius norm error to MSR-CV: error less than 1.25 that of MSR-CV almost alwaysNo clear effect
use multivariate square-root lasso instead of MRCE-ApproxMSR-CV · Model 2 and 3, L1 penaltysquared Frobenius norm error: MSR-CV outperforms MRCE-Approx in nearly every setting where the precision matrix is nonsparseImproved
use MSR-CV instead of MRCE-Approx under group lassoMSR-CV · Model 1-3, group lasso, M2 betasquared Frobenius norm error: MSR-CV outperforms MRCE-Approx in every settingImproved
use MSR-CV instead of MRCE-Approx on Model 1 xi=0.3MSR-CV · Model 1 xi=0.3, normal errorssquared Frobenius norm error: MRCE-Approx slightly outperforms MSR-CV at xi=0.3Hurt
use ADMM instead of CVX solverMSR with L1 penalty · Model 1 normal errorsaverage computing time (s): 0.61 vs 82.89 seconds at xi=0.3Improved
use ADMM instead of MRCE-ApproxMSR-CV · Model 1 normal errorsimplementation time: MSR-CV under a minute versus MRCE-Approx over 250 secondsImproved
use MSR-CV instead of PLS/Calibrated on Model 2 with small nMSR-CV · Model 2, (n,p,q)=(50,500,60)squared Frobenius norm error: MSR-CV and MRCE-Approx worse than PLS and CalibratedHurt
use theory-based lambda (MSR-q95) instead of cross-validation (ablation)MSR · Model 1 xi=0.3, L1 penaltytrue positive rate: 0.812 -> 0.487Hurt
use theory-based lambda (MSR-q95) instead of cross-validation (ablation)MSR · Model 1 xi=0.3, L1 penaltyfalse positive rate: 0.043 -> 0.000Improved
use theory-based lambda (MSR-q95) instead of cross-validation (ablation)MSR · Model 1 xi=0.3, nuclear normaverage estimated rank: 14.860 -> 2.590Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy