AI Paper Atlas

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

Exploring and Interacting with the Set of Good Sparse Generalized Additive Models

Chudi Zhong, Zhi Chen, Jiachang Liu, Margo Seltzer, Cynthia Rudin

NeurIPS 2023 · Learning theory and algorithms › Prediction and decision losses · approach: Active learning and subset selection · 4 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 blocking (Method 2) instead of re-optimization (Method 1) for a new support setRashomon set approximation · COMPAS K~=15time (s): 324.85 -> 4.84e-4Improved
use blocking (Method 2) instead of re-optimization (Method 1) for a new support setRashomon set approximation · MIMIC-II K~=35time (s): 1127.10 -> 5.89e-4Improved
use blocking (Method 2) instead of Method 1Rashomon set approximation · FICO K~=28precision ratio: 1.01No clear effect
shrink support set from K~=15 to 13 binsMethod 2 · COMPASvolume ratio: 1.05 -> 1.20Hurt
shrink support set from 28 to 24 binsMethod 2 · FICOvolume ratio: 0.90 -> 1.29Hurt
optimize log-determinant convex objective instead of original ellipsoid objective (ablation)Method 1 · COMPAS/FICO/MIMIC-IIprecision ratio: 1.01 -> 1.00No clear effect
solve LP with sign constraint instead of MIP for VI+ (fix other coefs)variable importance upper bound · Diabetestime (s): 18.876 -> 4.780Improved
solve LP with sign constraint instead of MIP for VI+ (not fix other coefs)variable importance upper bound · COMPAStime (s): 46.782 -> 3.043Improved
use proposed ellipsoid method instead of EBM+samplingRashomon set approximation · FICO lambda_s=5e-4running time (s): 1169.73 -> 131.71Improved
use proposed ellipsoid method instead of logistic regression+samplingRashomon set approximation · COMPAS lambda_s=5e-4running time (s): 22.65 -> 94.92Hurt
use proposed ellipsoid method instead of logistic regression+samplingRashomon set approximation · MIMIC-II lambda_s=5e-4running time (s): 786.85 -> 437.87Improved
increase theta from 1.005L* to 1.1L*ellipsoid scaling needed · Diabetesscaling ratio: 1.081 -> 1.119Hurt

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