AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Optimal Parameter-Transfer Learning by Semiparametric Model Averaging

Xiaonan Hu, Xinyu Zhang

JMLR 2023 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Statistical transfer learning

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 semiparametric model averaging over source models instead of target-only least squaresTrans-SMAP · Simulation, correct target model, M=3, n0=150averaged out-of-sample MSE: 0.030 -> 0.027Improved
borrow all source data via pooled least squares instead of averaging weightsLSE-All · Simulation, correct target model, M=3, n0=150averaged out-of-sample MSE: 0.027 -> 0.347Hurt
replace simple averaging weights with the proposed weight criterionTrans-SimpMA · Simulation, correct target model, M=3, n0=150averaged out-of-sample MSE: 0.239 -> 0.027Improved
use proposed weight criterion instead of BIC weightsTrans-SBIC · Simulation, correct target model, M=3, n0=150averaged out-of-sample MSE: 0.188 -> 0.027Improved
use proposed weight criterion instead of AIC weightsTrans-SAIC · Simulation, correct target model, M=3, n0=150averaged out-of-sample MSE: 0.183 -> 0.027Improved
use semiparametric transfer instead of parametric Trans-LassoTrans-SMAP · Simulation, correct target model, M=3, n0=150averaged out-of-sample MSE: 0.123 -> 0.027Improved
use semiparametric transfer instead of parametric Trans-GLMTrans-SMAP · Simulation, correct target model, M=6, n0=150averaged out-of-sample MSE: 0.127 -> 0.026Improved
misspecified target model, compare to LSE-TarTrans-SMAP · Simulation, misspecified target model, M=3, n0=150averaged out-of-sample MSE: 0.038 -> 0.034Improved
increase target sample size n0 from 150 to 500Trans-SMAP · Simulation, correct target model, M=3averaged out-of-sample MSE: 0.027 -> 0.008Improved
increase number of candidate models from M=3 to M=6Trans-SMAP · Simulation, correct target model, n0=150averaged out-of-sample MSE: 0.027 -> 0.026No clear effect
vary cross-validation folds J (2, 5, 10, leave-one-out) in weight choice (ablation)Trans-SMAP · Beijing housing rental dataaveraged out-of-sample MSENo clear effect
use Trans-SMAP instead of Trans-SimpMA, Trans-SBIC, Trans-SAICTrans-SMAP · Beijing housing rental data (MSPE)averaged out-of-sample MSEImproved

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