mlsanityData to 14 Oct 2026

Unsupervised Discovery of Interpretable Directions in the GAN Latent Space

arXiv 2002.03754 ↗First posted 10 Feb 2020cs.LG cs.CV stat.MLcited by 289

Not on the map. Why?

Cited by 24 in Diffusion models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

The latent spaces of GAN models often have semantically meaningful directions. Moving in these directions corresponds to human-interpretable image transformations, such as zooming or recoloring, enabling a more controllable generation process. However, the discovery of such directions is currently performed in a supervised manner, requiring human labels, pretrained models, or some form of self-supervision. These requirements severely restrict a range of directions existing approaches can discover. In this paper, we introduce an unsupervised method to identify interpretable directions in the latent space of a pretrained GAN model. By a simple model-agnostic procedure, we find directions corresponding to sensible semantic manipulations without any form of (self-)supervision. Furthermore, we reveal several non-trivial findings, which would be difficult to obtain by existing methods, e.g., a direction corresponding to background removal. As an immediate practical benefit of our work, we show how to exploit this finding to achieve competitive performance for weakly-supervised saliency detection.

Key facts. Cited by 289 papers in the map (data through 14 Oct 2026). First posted on arXiv 10 Feb 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 66; 2022: 88; 2023: 65; 2024: 42; 2025: 16; 2026: 12.

289 papers.

Sort:

By area:

  1. 2021StyleCLIP: Text-Driven Manipulation of StyleGAN Imagerycited by 991
  2. 2021Designing an Encoder for StyleGAN Image Manipulationcited by 544
  3. 2021Label-Efficient Semantic Segmentation with Diffusion Modelscited by 499
  4. 2022StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasetscited by 492
  5. 2021Pivotal Tuning for Latent-based Editing of Real Imagescited by 427
  6. 2021GAN Inversion: A Surveycited by 367
  7. 2022From Anecdotal Evidence to Quantitative Evaluation Methods: A Systematic Review on Evaluating Explainable AIcited by 302
  8. 2021ReStyle: A Residual-Based StyleGAN Encoder via Iterative Refinementcited by 248
  9. 2021HyperStyle: StyleGAN Inversion with HyperNetworks for Real Image Editingcited by 208
  10. 2021High-Fidelity GAN Inversion for Image Attribute Editingcited by 197
  11. 2022Latent Image Animator: Learning to Animate Images via Latent Space Navigationcited by 175
  12. 2021A Good Image Generator Is What You Need for High-Resolution Video Synthesiscited by 163
38 more
  1. 2021EditGAN: High-Precision Semantic Image Editingcited by 156
  2. 2021Interpretable Deep Learning: Interpretation, Interpretability, Trustworthiness, and Beyondcited by 118
  3. 2022Disentangled Representation Learningcited by 112
  4. 2021Counterfactual Generative Networkscited by 111
  5. 2021Talk-to-Edit: Fine-Grained Facial Editing via Dialogcited by 102
  6. 2021Only a Matter of Style: Age Transformation Using a Style-Based Regression Modelcited by 92
  7. 2021Exploiting Spatial Dimensions of Latent in GAN for Real-time Image Editingcited by 91
  8. 2021Repurposing GANs for One-shot Semantic Part Segmentationcited by 88
  9. 2021HyperInverter: Improving StyleGAN Inversion via Hypernetworkcited by 85
  10. 2021CLIP2StyleGAN: Unsupervised Extraction of StyleGAN Edit Directionscited by 78
  11. 2022Stitch it in Time: GAN-Based Facial Editing of Real Videoscited by 68
  12. 2022Unsupervised Salient Object Detection with Spectral Cluster Votingcited by 56
  13. 2021Towards Vivid and Diverse Image Colorization with Generative Color Priorcited by 55
  14. 2022Third Time's the Charm? Image and Video Editing with StyleGAN3cited by 54
  15. 2022Unsupervised Representation Learning from Pre-trained Diffusion Probabilistic Modelscited by 54
  16. 2024Diffusion Models and Representation Learning: A Surveycited by 54
  17. 2021Enjoy Your Editing: Controllable GANs for Image Editing via Latent Space Navigationcited by 53
  18. 2021Ensembling with Deep Generative Viewscited by 53
  19. 2021Low-Rank Subspaces in GANscited by 53
  20. 2021LatentCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directionscited by 52
  21. 2025From Flat to Hierarchical: Extracting Sparse Representations with Matching Pursuitcited by 49
  22. 2021WarpedGANSpace: Finding non-linear RBF paths in GAN latent spacecited by 48
  23. 2023HyperReenact: One-Shot Reenactment via Jointly Learning to Refine and Retarget Facescited by 48
  24. 2023NoiseCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions in Diffusion Modelscited by 47
  25. 2021Discover the Unknown Biased Attribute of an Image Classifiercited by 46
  26. 2021Finding an Unsupervised Image Segmenter in Each of Your Deep Generative Modelscited by 44
  27. 2023StyleGAN knows Normal, Depth, Albedo, and Morecited by 44
  28. 2021StyleMC: Multi-Channel Based Fast Text-Guided Image Generation and Manipulationcited by 40
  29. 2021Image-Based CLIP-Guided Essence Transfercited by 39
  30. 2022Spatially-Adaptive Multilayer Selection for GAN Inversion and Editingcited by 39
  31. 2022Self-Distilled StyleGAN: Towards Generation from Internet Photoscited by 37
  32. 2022Style Transformer for Image Inversion and Editingcited by 37
  33. 2024Interpreting the Weight Space of Customized Diffusion Modelscited by 36
  34. 2021Orthogonal Jacobian Regularization for Unsupervised Disentanglement in Image Generationcited by 35
  35. 2022Text and Image Guided 3D Avatar Generation and Manipulationcited by 35
  36. 2021Learning Disentangled Representation by Exploiting Pretrained Generative Models: A Contrastive Learning Viewcited by 34
  37. 2021EigenGAN: Layer-Wise Eigen-Learning for GANscited by 34
  38. 2021Predict, Prevent, and Evaluate: Disentangled Text-Driven Image Manipulation Empowered by Pre-Trained Vision-Language Modelcited by 34

50 most cited of 289 All 289 (JSON)