AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Fréchet Random Forests for Metric Space Valued Regression with Non Euclidean Predictors

Louis Capitaine, Jérémie Bigot, Rodolphe Thiébaut, Robin Genuer

JMLR 2024 · Learning theory and algorithms › Prediction and decision losses · approach: Approximate nearest neighbor indexing

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
replace Frechet tree by Frechet random forestFrechet tree · Simulation scenario 1, 2 behavior functions, n=100MSE: 0.047 -> 0.028Improved
use Frechet RF instead of FDboostFDboost · Simulation scenario 1, 2 behavior functions, n=1000MSE: 0.031 -> 0.006Improved
use Frechet RF instead of FDboostFDboost · Simulation scenario 1, 2 behavior functions, n=100MSE: 0.05 -> 0.028Improved
use Frechet RF instead of standard random forest on the air quality datastandard RF · Air quality CO predictionhourly MSE: 1.486 -> 1.160Improved
increase mtry in Frechet RF (ablation)Frechet RF · Simulation scenario 1, n=100MSEImproved
use extremely randomized Frechet RF with ntry=3 instead of full Frechet RF (ablation)Frechet RF · Simulation scenario 1, n=100build time: 281 seconds -> 191 secondsImproved
increase ntry in extremely randomized Frechet RF (ablation)ERFRF · Simulation scenario 1, n=100MSEImproved
remove 10%, 20%, 30% of observation times (ablation)Frechet RF · Simulation scenario 1, n=100MSEHurt
apply a constant time shift of 1 to output curves (ablation)Frechet RF · Simulation scenario 1, n=100Frechet-distance errorNo clear effect
apply random time shifts U([0,0.5]) to output curves (ablation)Frechet RF · Simulation scenario 1, n=100Frechet-distance error: increases slightly to 5.2 (FDboost without shifts 11.4)Hurt
add image variables to scalar variables as inputs (ablation)ERFRF · Scenario 2, n=100OOB errorImproved

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