AI Paper Atlas

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

A Sober Look at the Unsupervised Learning of Disentangled Representations and their Evaluation

Francesco Locatello, Stefan Bauer, Mario Lucic, Gunnar Raetsch, Sylvain Gelly, Bernhard Schölkopf, Olivier Bachem

JMLR 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 37 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
add regularization strength to the objective function as predictor of disentanglement scores (ablation)OLS variance analysis · all data setsvariance explained (%): 37% -> 59%Improved
increase beta regularization strengthbeta-VAE and beta-TCVAE · total correlation of mean representationtotal correlation: increased regularization leads to substantially increased total correlation of mean representationHurt
increase beta regularization strengthbeta-VAE and beta-TCVAE · total correlation of sampled representationtotal correlation: decreases on all data sets as beta increasedImproved
add regularization strength in FactorVAEFactorVAE · total correlation of sampled representationtotal correlation: no decrease with increasing regularizationNo clear effect
choose model by the same metric and data set (different seed) instead of random selectionhyperparameter selection · disentanglement metricsprobability of outperforming random: 52.7% -> 81.9%Improved
transfer hyperparameters across data sets for the same metrichyperparameter selection · disentanglement metricsprobability of outperforming random: 81.9% -> 59.6%Mixed
use unsupervised scores (reconstruction, ELBO, total correlation) for model selectionmodel selection · disentanglement metricsrank correlationNo clear effect
choose a different objective function (model)unsupervised disentanglement models · disentanglement scoresdisentanglement scoreNo clear effect
transfer hyperparameters from dSprites to Color-dSpriteshyperparameter transfer · Color-dSpritesrank correlationImproved

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