AI Paper Atlas

Research map / Learning theory and algorithms / Submodular and game algorithms

Responsible AI (RAI) Games and Ensembles

Yash Gupta, Runtian Zhai, Arun Suggala, Pradeep Kumar Ravikumar

NeurIPS 2023 · Learning theory and algorithms › Submodular and game algorithms · approach: Opponent modeling and subgame solving · 1 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
replace ERM with RAI-FW (chi2) ensemble, domain oblivioussmall base learners · CIFAR10worst-class loss: 27.0 -> 23.4Improved
replace ERM with RAI-FW (chi2) ensemble, domain oblivioussmall base learners · CIFAR100worst-class loss: 66.0 -> 63.8Improved
replace Online GDRO with RAI-FW (Group), domain awaresmall base learners · CIFAR100worst-class loss: 57.0 -> 52.9Improved
replace Online GDRO with RAI-FW (Group), domain awaresmall base learners · CIFAR10worst-class loss: 21.7 -> 20.3Improved
replace Online GDRO with RAI-FW (Group and chi2), partial domain awaresmall base learners · CIFAR100worst-class loss: 64.0 -> 58.9Improved
replace Online GDRO with RAI-GA (Group and chi2), partial domain awaresmall base learners · COMPASworst-group loss: 32.7 -> 32.9No clear effect
use AdaBoost instead of ERMsmall base learners · CIFAR-10 (Imbalanced)worst-class loss: 33.6 -> 40.6Hurt
use RAI-FW (chi2) ensemble instead of ERM, average-case checksmall base learners · CIFAR100average loss: 27.1 -> 27.5No clear effect
use RAI-FW (Group) instead of Online GDRO, average-case checksmall base learners · COMPASaverage loss: 31.7 -> 32.1No clear effect
use complex larger base models with RAI-GA instead of ERM complexcomplex models · CIFAR10worst-class loss: 21.3 -> 15.4Improved
use larger complex models instead of small models with RAI ensembleRAI-FW (chi2) · CIFAR10worst-class loss: 23.4 -> 15.4Improved
use RAI-GA instead of RAI-FW (chi2) on small DOsmall base learners · CIFAR10worst-class loss: 25.0 -> 23.4No clear effect

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