AI Paper Atlas

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

Adapting and Evaluating Influence-Estimation Methods for Gradient-Boosted Decision Trees

Jonathan Brophy, Zayd Hammoudeh, Daniel Lowd

JMLR 2023 · Learning theory and algorithms › Prediction and decision losses · approach: Tree ensemble optimization · 12 citations (OpenAlex)

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 GBDT models instead of LR/DT/KNN/SVM/RF/MLPtabular prediction · 22 tabular data setstest AUC/accuracy/MSE: Concrete MSE 50.6 -> 18.8Improved
use BoostIn instead of Random for multi-test removalinfluence estimation for GBDT · multiple test instances, removing examplesloss increase relative to random: 2.2xImproved
use BoostIn instead of LOO for removing a group of examplesinfluence estimation for GBDT · multiple test instances, removing examplesaverage rankImproved
use LeafRefit instead of BoostIn for single-test removalinfluence estimation for GBDT · single test instance, removalaverage rankImproved
use LeafRefit/TREX/TreeSim instead of BoostIn for targeted label editsinfluence estimation for GBDT · single test instance, targeted label editsaverage rankImproved
add fixed-structure assumption (LeafRefit) to LOO (ablation)LOO · single test instance, removaltest loss increaseImproved
reestimate influence after each removal instead of fixed ordering (ablation)all influence methods · one-at-a-time removal, LGBtest loss increaseImproved
use LeafInfluence/LeafRefit instead of LOO for setup timeinfluence estimation · runtime, SDS data setssetup timeHurt
use BoostIn instead of Loss baseline for fixing mislabelled examplesinfluence estimation · fix mislabelled, 40% flipped labelsaverage rankImproved
use BoostIn instead of LeafInfSP for adding label noiseinfluence estimation · multiple test instances, label noiseaverage rankImproved

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