Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
119 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Incremental Generalized Category Discovery (2304.14310v3)

Published 27 Apr 2023 in cs.CV

Abstract: We explore the problem of Incremental Generalized Category Discovery (IGCD). This is a challenging category incremental learning setting where the goal is to develop models that can correctly categorize images from previously seen categories, in addition to discovering novel ones. Learning is performed over a series of time steps where the model obtains new labeled and unlabeled data, and discards old data, at each iteration. The difficulty of the problem is compounded in our generalized setting as the unlabeled data can contain images from categories that may or may not have been observed before. We present a new method for IGCD which combines non-parametric categorization with efficient image sampling to mitigate catastrophic forgetting. To quantify performance, we propose a new benchmark dataset named iNatIGCD that is motivated by a real-world fine-grained visual categorization task. In our experiments we outperform existing related methods

Definition Search Book Streamline Icon: https://streamlinehq.com
References (66)
  1. Human category learning. Annual review of psychology, 2005.
  2. Masked siamese networks for label-efficient learning. In ECCV, 2022.
  3. Semi-supervised learning of visual features by non-parametrically predicting view assignments with support samples. In ICCV, 2021.
  4. Towards open world recognition. In CVPR, 2015.
  5. Towards open set deep networks. In CVPR, 2016.
  6. Mixmatch: A holistic approach to semi-supervised learning. In NeurIPS, 2019.
  7. Learning and the unknown: Surveying steps toward open world recognition. In AAAI, 2019.
  8. Dark experience for general continual learning: a strong, simple baseline. In NeurIPS, 2020.
  9. Open-world semi-supervised learning. In ICLR, 2022.
  10. Emerging properties in self-supervised vision transformers. In ICCV, 2021.
  11. Semi-Supervised Learning. The MIT Press, 2006.
  12. A simple framework for contrastive learning of visual representations. In ICML, 2020.
  13. Taxonomic class incremental learning. arXiv:2304.05547, 2023.
  14. Parametric information maximization for generalized category discovery. arXiv:2212.00334, 2022.
  15. An image is worth 16x16 words: Transformers for image recognition at scale. ICLR, 2021.
  16. Xcon: Learning with experts for fine-grained category discovery. In BMVC, 2022.
  17. A unified objective for novel class discovery. In ICCV, 2021.
  18. Recent advances in open set recognition: A survey. PAMI, 2020.
  19. Automatically discovering and learning new visual categories with ranking statistics. In ICLR, 2020.
  20. Learning to discover novel visual categories via deep transfer clustering. In ICCV, 2019.
  21. Momentum contrast for unsupervised visual representation learning. In CVPR, 2020.
  22. Deep residual learning for image recognition. In CVPR, 2016.
  23. Learning to cluster in order to transfer across domains and tasks. In ICLR, 2018.
  24. Multi-class classification without multi-class labels. In ICLR, 2019.
  25. Trash to treasure: harvesting ood data with cross-modal matching for open-set semi-supervised learning. In ICCV, 2021.
  26. iNaturalist. iNaturalist. www.inaturalist.org, accessed 8 March 2023.
  27. Novel class discovery without forgetting. In ECCV, 2022.
  28. Geonet: Benchmarking unsupervised adaptation across geographies. In CVPR, 2023.
  29. Supervised contrastive learning. In NeurIPS, 2020.
  30. Overcoming catastrophic forgetting in neural networks. PNAS, 2017.
  31. 3d object representations for fine-grained categorization. In Workshop on 3D Representation and Recognition, 2013.
  32. Alex Krizhevsky. Learning multiple layers of features from tiny images. University of Toronto, 2009.
  33. Temporal ensembling for semi-supervised learning. In ICLR, 2017.
  34. Human-level concept learning through probabilistic program induction. Science, 2015.
  35. Density-aware graph for deep semi-supervised visual recognition. In CVPR, 2020.
  36. Continual learning with evolving class ontologies. NeurIPS, 2022.
  37. Large-scale pre-trained models are surprisingly strong in incremental novel class discovery. arXiv:2303.15975, 2023.
  38. Residual tuning: Toward novel category discovery without labels. Transactions on Neural Networks and Learning Systems, 2022.
  39. Sgdr: Stochastic gradient descent with warm restarts. In ICLR, 2017.
  40. Fine-grained visual classification of aircraft. arXiv:1306.5151, 2013.
  41. Catastrophic interference in connectionist networks: The sequential learning problem. In Psychology of learning and motivation, 1989.
  42. A wholistic view of continual learning with deep neural networks: Forgotten lessons and the bridge to active and open world learning. Neural Networks, 2023.
  43. Unified probabilistic deep continual learning through generative replay and open set recognition. Journal of Imaging, 2022.
  44. Efficient non-maximum suppression. In ICPR, 2006.
  45. Realistic evaluation of deep semi-supervised learning algorithms. In NeurIPS, 2018.
  46. icarl: Incremental classifier and representation learning. In CVPR, 2017.
  47. In defense of pseudo-labeling: An uncertainty-aware pseudo-label selection framework for semi-supervised learning. In ICLR, 2021.
  48. Anthony Robins. Catastrophic forgetting, rehearsal and pseudorehearsal. Connection Science, 1995.
  49. Class-incremental novel class discovery. In ECCV, 2022.
  50. ImageNet Large Scale Visual Recognition Challenge. IJCV, 2015.
  51. Openmatch: Open-set consistency regularization for semi-supervised learning with outliers. In NeurIPS, 2021.
  52. Toward open set recognition. PAMI, 2012.
  53. Fixmatch: Simplifying semi-supervised learning with consistency and confidence. In NeurIPS, 2020.
  54. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. In NeurIPS, 2017.
  55. Sebastian Thrun. Is learning the n-th thing any easier than learning the first? In NeurIPS, 1995.
  56. Novel class discovery: an introduction and key concepts. arXiv:2302.12028, 2023.
  57. Benchmarking representation learning for natural world image collections. In CVPR, 2021.
  58. Generalized category discovery. In CVPR, 2022.
  59. Open-set recognition: a good closed-set classifier is all you need? In ICLR, 2022.
  60. Caltech-ucsd birds 200. Technical Report CNS-TR-2011-001, California Institute of Technology, 2011.
  61. Parametric classification for generalized category discovery: A baseline study. In ICCV, 2023.
  62. Multi-task curriculum framework for open-set semi-supervised learning. In ECCV, 2020.
  63. Grow and merge: A unified framework for continuous categories discovery. In NeurIPS, 2022.
  64. Novel visual category discovery with dual ranking statistics and mutual knowledge distillation. In NeurIPS, 2021.
  65. Deep class-incremental learning: A survey. arXiv:2302.03648, 2023.
  66. Xiaojin Jerry Zhu. Semi-supervised learning literature survey. University of Wisconsin-Madison Department of Computer Sciences, 2005.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (2)
  1. Bingchen Zhao (47 papers)
  2. Oisin Mac Aodha (62 papers)
Citations (15)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now