AI Paper Atlas

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

Learning Models for Actionable Recourse

Alexis Ross, Himabindu Lakkaraju, Osbert Bastani

NeurIPS 2021 · Learning theory and algorithms › Prediction and decision losses · approach: Counterfactual explanations and recourse · 0 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
Train with the recourse loss (λ=0.8) instead of the baseline lossNeural network classifier · Adult, Compas, Bail, Germanrecourse neg using adversarial training: 0.007 -> 0.498; 0.000 -> 0.967; 0.000 -> 0.993; 0.127 -> 0.968Improved
Train with the recourse loss (λ=0.8) instead of the baseline lossNeural network classifier · Adult, Compas, Bail, GermanF1 score: 0.697 -> 0.636; 0.739 -> 0.717; 0.775 -> 0.760; 0.447 -> 0.419Hurt
Train with the recourse loss (λ=0.8) instead of the baseline lossNeural network classifier · Adult, Compas, Bail, Germanrecourse neg using gradient descent: 0.210 -> 0.496; 0.579 -> 1.000; 0.317 -> 1.000; 0.804 -> 0.864Improved
Train with the recourse loss (λ=0.8) instead of the baseline lossNeural network classifier · Adult, Compas, Bail, Germanrecourse neg using linear approximation: 0.220 -> 0.053; 0.156 -> 0.068; 0.102 -> 0.029; 0.204 -> 0.086Hurt
Use the adversarial-training recourse solver rather than gradient descentModels trained with the recourse loss · Adult, Compas, Bail, Germanrecourse neg: 0.498 vs 0.496; 0.967 vs 1.000; 0.993 vs 1.000; 0.968 vs 0.864Mixed
Choose the threshold to satisfy PARE instead of maximizing F1Baseline classifier · AdultF1 score: 0.697 -> 0.400Hurt
Choose the threshold to satisfy PARE instead of maximizing F1Recourse-trained classifier · AdultF1 score: 0.636 -> 0.526Hurt
Choose the threshold to satisfy PARE instead of maximizing F1Recourse-trained classifier · Compas, Bail, GermanF1 score: 0.717 -> 0.721; 0.760 -> 0.776; 0.419 -> 0.457No clear effect
Vary recourse-loss weight λ from zero to positive values (ablation)Neural network classifier · Adult, Compas, Bail, Germanrecourse rateImproved
Use linear approximation instead of gradient descent or adversarial training to compute recourse (ablation)Recourse-trained classifier · Adult, Compas, Bail, Germanrecourse rate: 0.053 vs 0.496; 0.068 vs 1.000; 0.029 vs 1.000; 0.086 vs 0.864Hurt
Use recourse loss (λ=0.25) instead of baseline loss (λ=0)Sentiment classifier · Stanford Sentiment Treebank test setrecourse neg: 0.239 -> 0.950Improved
Use recourse loss (λ=0.25) instead of baseline loss (λ=0)Sentiment classifier · Stanford Sentiment Treebank test setF1 score: 0.881 -> 0.864Hurt

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