AI Paper Atlas

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

Using Shapley Values and Variational Autoencoders to Explain Predictive Models with Dependent Mixed Features

Lars H. B. Olsen, Ingrid K. Glad, Martin Jullum, Kjersti Aas

JMLR 2022 · Learning theory and algorithms › Prediction and decision losses · approach: Shapley value attribution · 5 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 VAEAC instead of Gaussian approachShapley value estimation · Burr continuous, M=5, N_train=1000EC1: 0.0424 -> 0.0264Improved
increase training size from 1000 to 5000VAEAC · Burr continuous, M=10EC1: 0.0288 -> 0.0218Improved
increase training size from 1000 to 5000Independence approach · Burr continuous, M=10EC1: 0.1082 -> 0.1160Hurt
increase training size from 1000 to 5000Gaussian approach · Burr continuous, M=10EC1: 0.0384 -> 0.0384No clear effect
use coalition-focused masking (VAEAC_C) instead of uniform maskingVAEAC · Burr continuous, M=100, N_train=10000EC1: 0.1013 -> 0.0504Improved
use coalition-focused masking (VAEAC_C) instead of uniform maskingVAEAC · Burr continuous, M=250, N_train=25000EC1: 0.6891 -> 0.1474Improved
use coalition-focused masking (VAEAC_C) instead of uniform maskingVAEAC · Burr continuous, M=25, N_train=1000EC1: 0.0321 -> 0.0346Hurt
use VAEAC_C instead of Gaussian at limited training sizeShapley estimation · Burr continuous, M=100, N_train=5000EC1: 0.0555 -> 0.0594Hurt
use VAEAC instead of independence under strong dependenceShapley estimation · Mixed M=4, N_train=5000, rho=0.9EC1: 0.6464 -> 0.0489Improved
use VAEAC instead of independence for independent featuresShapley estimation · Mixed M=4, N_train=5000, rho=0.0EC1: 0.0240 -> 0.0523Hurt
use VAEAC instead of ctree with moderate training setShapley estimation · Mixed M=4, N_train=1000, rho=0.1EC1: 0.0706 -> 0.0770Hurt
use VAEAC instead of ctree for larger categorical dimensionShapley estimation · Categorical M=7 L=4, rho=0.9EC1: 0.0736 -> 0.0647Improved

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