mlsanityData to 7 Oct 2026

Evaluating Disentanglement of Structured Representations

arXiv 2101.04041 ↗First posted 11 Jan 2021cs.LG cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

We introduce the first metric for evaluating disentanglement at individual hierarchy levels of a structured latent representation. Applied to object-centric generative models, this offers a systematic, unified approach to evaluating (i) object separation between latent slots (ii) disentanglement of object properties inside individual slots (iii) disentanglement of intrinsic and extrinsic object properties. We theoretically show that for structured representations, our framework gives stronger guarantees of selecting a good model than previous disentanglement metrics. Experimentally, we demonstrate that viewing object compositionality as a disentanglement problem addresses several issues with prior visual metrics of object separation. As a core technical component, we present the first representation probing algorithm handling slot permutation invariance.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 24 of its 40 references that are in the map. First posted on arXiv 11 Jan 2021.

Builds on

24 of 40 references.

References · 24

  1. 2020End-to-End Object Detection with Transformerscited by 7,384
  2. 2016CLEVR: A Diagnostic Dataset for Compositional Language and Elementary Visual Reasoningcited by 1,445
  3. 2018Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representationscited by 929
  4. 2020Object-Centric Learning with Slot Attentioncited by 865
  5. 2020GIRAFFE: Representing Scenes as Compositional Generative Neural Feature Fieldscited by 784
  6. 2018Disentangling by Factorisingcited by 614
  7. 2019CLEVRER: CoLlision Events for Video REpresentation and Reasoningcited by 475
  8. 2019MONet: Unsupervised Scene Decomposition and Representationcited by 388
  9. 2019Multi-Object Representation Learning with Iterative Variational Inferencecited by 359
  10. 2017Variational Inference of Disentangled Latent Concepts from Unlabeled Observationscited by 240
  11. 2019GENESIS: Generative Scene Inference and Sampling with Object-Centric Latent Representationscited by 227
  12. 2020SPACE: Unsupervised Object-Oriented Scene Representation via Spatial Attention and Decompositioncited by 190
12 more
  1. 2016Permutation Invariant Training of Deep Models for Speaker-Independent Multi-talker Speech Separationcited by 189
  2. 2018Learning Latent Permutations with Gumbel-Sinkhorn Networkscited by 172
  3. 2020BlockGAN: Learning 3D Object-aware Scene Representations from Unlabelled Imagescited by 166
  4. 2017Neural Expectation Maximizationcited by 123
  5. 2018Learning Deep Disentangled Embeddings with the F-Statistic Losscited by 101
  6. 2018Relational Neural Expectation Maximization: Unsupervised Discovery of Objects and their Interactionscited by 100
  7. 2016Tagger: Deep Unsupervised Perceptual Groupingcited by 69
  8. 2020Measuring Disentanglement: A Review of Metricscited by 69
  9. 2018Structured Disentangled Representationscited by 62
  10. 2020RELATE: Physically Plausible Multi-Object Scene Synthesis Using Structured Latent Spacescited by 32
  11. 2020Slot Contrastive Networks: A Contrastive Approach for Representing Objectscited by 9
  12. 2018Investigating Object Compositionality in Generative Adversarial Networkscited by 3

Built on by

Cited by year

Citing papers per year: 2021: 1; 2022: 1; 2023: 0; 2024: 1.

3 papers.

  1. 2021Learning Disentangled Representations in the Imaging Domaincited by 46
  2. 2022Neural Systematic Bindercited by 37
  3. 2024Parallelized Spatiotemporal Bindingcited by 2