AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Controlling Directions Orthogonal to a Classifier

Yilun Xu, Hao He, Tianxiao Shen, Tommi S. Jaakkola

ICLR 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Statistical alignment and hypothesis testing · 3 citations (OpenAlex)

Read the paper

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
Add classifier orthogonalization to CycleGAN using the Fish principal classifierCycleGAN · CMNISTZ2 accuracy: 14.3 -> 98.0Improved
Add classifier orthogonalization using an oracle principal classifierCycleGAN · CelebA-GHZ2 accuracy: 16.8 -> 57.4Improved
Add classifier orthogonalization using an oracle principal classifierCycleGAN · CelebA-GHFID: 38.5 -> 39.7Hurt
Orthogonalize the domain discriminator with respect to task labelsVADA · Visual domain adaptation with label shifttest accuracy: VADA + classifier orthogonalization improves over VADA across all tasksImproved
Use importance-weighted domain adaptation rather than the original VADA objectiveVADA · MNIST → MNIST-M and MNIST → SVHNtest accuracy: 77.8 -> 71.2; 35.7 -> 34.5Hurt
Orthogonalize the ERM classifier with respect to genderERM classifier · UCI AdultΔEO: 0.20 -> 0.12Improved
Orthogonalize the ERM classifier with respect to genderERM classifier · UCI Adultaccuracy: 84.5 -> 81.6Hurt
Orthogonalize the ERM classifier with respect to ageERM classifier · UCI German creditΔDP: 0.19 -> 0.09Improved
Orthogonalize the ERM classifier with respect to ageERM classifier · UCI German creditΔEO: 0.33 -> 0.18Improved
Orthogonalize the ERM classifier with respect to ageERM classifier · UCI German creditaccuracy: 76.0 -> 75.4Hurt
Train the orthogonal classifier by importance sampling instead of classifier orthogonalizationOrthogonal classifier · Synthetic variation in latent divergence and learning ratetest lossHurt

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