AI Paper Atlas

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

Implications of Model Indeterminacy for Explanations of Automated Decisions

Marc-Etienne Brunet, Ashton Anderson, Richard Zemel

NeurIPS 2022 · 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
source of indeterminacy: Rashomon set instead of underspecificationDinput Shapley explanations · UCI Credit Cardtop-5 CDC (%): 9.3 -> 54.9Hurt
source of indeterminacy: Rashomon set instead of underspecificationDinput Shapley explanations · UCI Credit Cardtop-3 SSD (%): 51.1 -> 79.4Hurt
source of indeterminacy: Rashomon set instead of underspecificationDinput Shapley explanations · Porto Seguro Safe Drivertop-5 CDC (%): 0.6 -> 23.6Hurt
explain top-5 instead of top-3 featuresDinput Shapley, underspecification · UCI Credit CardSSD (%): 51.1 -> 67.2Hurt
explain top-5 instead of top-3 featuresDinput Shapley, Rashomon · Give Me Some CreditCDC (%): 11.1 -> 29.0Hurt
Dmean Shapley instead of Dinput ShapleyRashomon sets · UCI Credit Cardtop-3 CDC (%): 36.5 -> 46.6Hurt
Dmean Shapley instead of Dinput Shapleyunderspecification sets · UCI Credit Cardtop-3 SSD (%): 51.1 -> 72.1Hurt
two approximately equivalent models at 50% acceptance thresholdclassifiers · three credit datasetsdecision disagreement (%): 4.35-12.7Hurt
use epistemic uncertainty of the prediction to predict explanation consistencyapproximately equivalent models · UCI Credit Card and Safe DriverSpearman correlation with SA: -0.17 to 0.23No clear effect
use epistemic uncertainty of the prediction to predict explanation consistencyapproximately equivalent models · Give Me Some CreditSpearman correlation with SA: -0.34 to -0.58Mixed

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