AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Weakly-Supervised Disentanglement Without Compromises

Francesco Locatello, Ben Poole, Gunnar Raetsch, Bernhard Schölkopf, Olivier Bachem, Michael Tschannen

ICML 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 90 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
Train on pairs of observations with unknown shared factors using adaptive group inferenceDisentangled representation learning · dSprites, SmallNORB, Cars3D, Shapes3D, MPI3DDCI Disentanglement: 18.6% -> 47.9%; 28.0% -> 33.8%; 22.5% -> 54.0%; 70.9% -> 94.6%; 19.5% -> 40.1%Improved
Make the changes between factors sparserAda-GVAE · MPI3DDCI DisentanglementImproved
Provide exact knowledge of which factor changed when k=1Weakly-supervised disentanglement methods · MPI3DDCI DisentanglementImproved
Use incomplete group labels as though only one factor changedGVAE · MPI3D and datasets with more factors of variationDCI DisentanglementHurt
Replace adaptive GVAE averaging with adaptive ML-VAE averaging (ablation)Ada-GVAE · dSpritesDCI Disentanglement: 47.9% -> 29.4%Hurt
Replace adaptive GVAE averaging with adaptive ML-VAE averaging (ablation)Ada-GVAE · SmallNORBDCI Disentanglement: 33.8% -> 34.1%Improved
Replace adaptive GVAE averaging with adaptive ML-VAE averaging (ablation)Ada-GVAE · Cars3DDCI Disentanglement: 54.0% -> 45.6%Hurt
Replace adaptive GVAE averaging with adaptive ML-VAE averaging (ablation)Ada-GVAE · Shapes3DDCI Disentanglement: 94.6% -> 94.0%Hurt
Replace adaptive GVAE averaging with adaptive ML-VAE averaging (ablation)Ada-GVAE · MPI3DDCI Disentanglement: 40.1% -> 28.5%Hurt

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