mlsanityData to 14 Oct 2026

Invariant Information Clustering for Unsupervised Image Classification and Segmentation

arXiv 1807.06653 ↗First posted 17 Jul 2018cs.CV cs.LGcited by 455

Not on the map. Why?

Abstract (arXiv)

We present a novel clustering objective that learns a neural network classifier from scratch, given only unlabelled data samples. The model discovers clusters that accurately match semantic classes, achieving state-of-the-art results in eight unsupervised clustering benchmarks spanning image classification and segmentation. These include STL10, an unsupervised variant of ImageNet, and CIFAR10, where we significantly beat the accuracy of our closest competitors by 6.6 and 9.5 absolute percentage points respectively. The method is not specialised to computer vision and operates on any paired dataset samples; in our experiments we use random transforms to obtain a pair from each image. The trained network directly outputs semantic labels, rather than high dimensional representations that need external processing to be usable for semantic clustering. The objective is simply to maximise mutual information between the class assignments of each pair. It is easy to implement and rigorously grounded in information theory, meaning we effortlessly avoid degenerate solutions that other clustering methods are susceptible to. In addition to the fully unsupervised mode, we also test two semi-supervised settings. The first achieves 88.8% accuracy on STL10 classification, setting a new global state-of-the-art over all existing methods (whether supervised, semi-supervised or unsupervised). The second shows robustness to 90% reductions in label coverage, of relevance to applications that wish to make use of small amounts of labels. github.com/xu-ji/IIC

Key facts. Cited by 455 papers in the map (data through 14 Oct 2026). First posted on arXiv 17 Jul 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 115; 2022: 96; 2023: 107; 2024: 66; 2025: 40; 2026: 31.

455 papers.

Sort:

By area:

  1. 2021DatasetGAN: Efficient Labeled Data Factory with Minimal Human Effortcited by 264
  2. 2022Unsupervised Semantic Segmentation by Distilling Feature Correspondencescited by 241
  3. 2021Localizing Objects with Self-Supervised Transformers and no Labelscited by 230
  4. 2022Bridging the Gap to Real-World Object-Centric Learningcited by 220
  5. 2021Self-Supervised Pretraining of 3D Features on any Point-Cloudcited by 217
  6. 2021A Large-Scale Study on Unsupervised Spatiotemporal Representation Learningcited by 214
  7. 2021Unsupervised Semantic Segmentation by Contrasting Object Mask Proposalscited by 190
  8. 2021PiCIE: Unsupervised Semantic Segmentation using Invariance and Equivariance in Clusteringcited by 163
  9. 2021A Unified Objective for Novel Class Discoverycited by 162
  10. 2021Efficient Visual Pretraining with Contrastive Detectioncited by 150
  11. 2022Deep Spectral Methods: A Surprisingly Strong Baseline for Unsupervised Semantic Segmentation and Localizationcited by 150
  12. 2021Efficient Self-supervised Vision Transformers for Representation Learningcited by 149
38 more
  1. 2021Semantic Segmentation with Generative Models: Semi-Supervised Learning and Strong Out-of-Domain Generalizationcited by 141
  2. 2022Rethinking Semantic Segmentation: A Prototype Viewcited by 140
  3. 2023Diffuse, Attend, and Segment: Unsupervised Zero-Shot Segmentation using Stable Diffusioncited by 131
  4. 2021Self-supervised Graph-level Representation Learning with Local and Global Structurecited by 125
  5. 2022ReCo: Retrieve and Co-segment for Zero-shot Transfercited by 117
  6. 2021AutoNovel: Automatically Discovering and Learning Novel Visual Categoriescited by 116
  7. 2021Efficient Training of Visual Transformers with Small Datasetscited by 108
  8. 2021SPICE: Semantic Pseudo-labeling for Image Clusteringcited by 104
  9. 2021SubTab: Subsetting Features of Tabular Data for Self-Supervised Representation Learningcited by 104
  10. 2022FreeSOLO: Learning to Segment Objects without Annotationscited by 102
  11. 2021Large-scale Unsupervised Semantic Segmentationcited by 101
  12. 2021Pixel Contrastive-Consistent Semi-Supervised Semantic Segmentationcited by 101
  13. 2021SiT: Self-supervised vIsion Transformercited by 99
  14. 2021Continual Semantic Segmentation via Repulsion-Attraction of Sparse and Disentangled Latent Representationscited by 96
  15. 2022Attracting and Dispersing: A Simple Approach for Source-free Domain Adaptationcited by 94
  16. 2021ClevrTex: A Texture-Rich Benchmark for Unsupervised Multi-Object Segmentationcited by 93
  17. 2022Self-Supervised Learning of Object Parts for Semantic Segmentationcited by 90
  18. 2021Divide and Contrast: Self-supervised Learning from Uncurated Datacited by 85
  19. 2022Perceptual Grouping in Contrastive Vision-Language Modelscited by 82
  20. 2021PixMatch: Unsupervised Domain Adaptation via Pixelwise Consistency Trainingcited by 80
  21. 2022Object discovery and representation networkscited by 80
  22. 2022Deep Clustering: A Comprehensive Surveycited by 78
  23. 2022A Comprehensive Survey on Deep Clustering: Taxonomy, Challenges, and Future Directionscited by 73
  24. 2022NeRF-SOS: Any-View Self-supervised Object Segmentation on Complex Scenescited by 71
  25. 2022Self-Supervised Visual Representation Learning with Semantic Groupingcited by 70
  26. 2021Clusterability as an Alternative to Anchor Points When Learning with Noisy Labelscited by 65
  27. 2021Clustering-friendly Representation Learning via Instance Discrimination and Feature Decorrelationcited by 65
  28. 2021Playable Video Generationcited by 64
  29. 2021Graph Contrastive Clusteringcited by 61
  30. 2023No Representation Rules Them All in Category Discoverycited by 61
  31. 2022Twin Contrastive Learning for Online Clusteringcited by 60
  32. 2022Visual Recognition with Deep Nearest Centroidscited by 59
  33. 2021Detecting Corrupted Labels Without Training a Model to Predictcited by 58
  34. 2021Learning Representation for Clustering via Prototype Scattering and Positive Samplingcited by 58
  35. 2021Unsupervised Part Discovery from Contrastive Reconstructioncited by 57
  36. 2021Joint Representation Learning and Novel Category Discovery on Single- and Multi-modal Datacited by 55
  37. 2023Source-Free Domain Adaptation with Frozen Multimodal Foundation Modelcited by 55
  38. 2022Discovering Object Masks with Transformers for Unsupervised Semantic Segmentationcited by 54

50 most cited of 455 All 455 (JSON)