Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
194 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
46 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Knowledge Transfer-Driven Few-Shot Class-Incremental Learning (2306.10942v1)

Published 19 Jun 2023 in cs.CV and cs.AI

Abstract: Few-shot class-incremental learning (FSCIL) aims to continually learn new classes using a few samples while not forgetting the old classes. The key of this task is effective knowledge transfer from the base session to the incremental sessions. Despite the advance of existing FSCIL methods, the proposed knowledge transfer learning schemes are sub-optimal due to the insufficient optimization for the model's plasticity. To address this issue, we propose a Random Episode Sampling and Augmentation (RESA) strategy that relies on diverse pseudo incremental tasks as agents to achieve the knowledge transfer. Concretely, RESA mimics the real incremental setting and constructs pseudo incremental tasks globally and locally, where the global pseudo incremental tasks are designed to coincide with the learning objective of FSCIL and the local pseudo incremental tasks are designed to improve the model's plasticity, respectively. Furthermore, to make convincing incremental predictions, we introduce a complementary model with a squared Euclidean-distance classifier as the auxiliary module, which couples with the widely used cosine classifier to form our whole architecture. By such a way, equipped with model decoupling strategy, we can maintain the model's stability while enhancing the model's plasticity. Extensive quantitative and qualitative experiments on three popular FSCIL benchmark datasets demonstrate that our proposed method, named Knowledge Transfer-driven Relation Complementation Network (KT-RCNet), outperforms almost all prior methods. More precisely, the average accuracy of our proposed KT-RCNet outperforms the second-best method by a margin of 5.26%, 3.49%, and 2.25% on miniImageNet, CIFAR100, and CUB200, respectively. Our code is available at https://github.com/YeZiLaiXi/KT-RCNet.git.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (51)
  1. Product recognition for unmanned vending machines. IEEE Transactions on Neural Networks and Learning Systems, pages 1–14, 2022.
  2. Picassonet: Searching adaptive architecture for efficient facial landmark localization. IEEE Transactions on Neural Networks and Learning Systems, pages 1–12, 2022.
  3. Dbcface: Towards pure convolutional neural network face detection. IEEE Transactions on Circuits and Systems for Video Technology, 32(4):1792–1804, 2021.
  4. Learning without forgetting. In ECCV, pages 614–629, 2016.
  5. Model behavior preserving for class-incremental learning. IEEE Transactions on Neural Networks and Learning Systems, pages 1–12, 2022.
  6. Memory replay with data compression for continual learning. In ICLR, 2022.
  7. Few-shot class-incremental learning. In CVPR, June 2020.
  8. Self-promoted prototype refinement for few-shot class-incremental learning. In CVPR, pages 6801–6810, June 2021.
  9. Few-shot incremental learning with continually evolved classifiers. In CVPR, pages 12455–12464, June 2021.
  10. Constrained few-shot class-incremental learning. In CVPR, pages 9057–9067, 2022.
  11. Metafscil: A meta-learning approach for few-shot class incremental learning. In CVPR, pages 14166–14175, June 2022.
  12. Matching networks for one shot learning. In NeurIPS, volume 29, 2016.
  13. Forward compatible few-shot class-incremental learning. In CVPR, pages 9046–9056, June 2022.
  14. Prototypical networks for few-shot learning. In NeurIPS, volume 30, 2017.
  15. Cad: Co-adapting discriminative features for improved few-shot classification. In CVPR, pages 14554–14563, June 2022.
  16. Learning to affiliate: Mutual centralized learning for few-shot classification. In CVPR, pages 14411–14420, June 2022.
  17. Model-agnostic meta-learning for fast adaptation of deep networks. In ICML, volume 70, pages 1126–1135, 2017.
  18. Meta-learning with latent embedding optimization. In ICLR, 2019.
  19. Learning to forget for meta-learning. In CVPR, pages 2379–2387, 2020.
  20. Melr: Meta-learning via modeling episode-level relationships for few-shot learning. In ICLR, 2021.
  21. Attentive weights generation for few shot learning via information maximization. In CVPR, June 2020.
  22. Improving adversarially robust few-shot image classification with generalizable representations. In CVPR, pages 9025–9034, June 2022.
  23. Generating representative samples for few-shot classification. In CVPR, pages 9003–9013, June 2022.
  24. Podnet: Pooled outputs distillation for small-tasks incremental learning. In ECCV, pages 86–102, 2020.
  25. Distilling causal effect of data in class-incremental learning. In CVPR, pages 3957–3966, June 2021.
  26. icarl: Incremental classifier and representation learning. In CVPR, July 2017.
  27. Learning bayesian sparse networks with full experience replay for continual learning. In CVPR, pages 109–118, June 2022.
  28. Experience replay for continual learning. In NeurIPS, volume 32, 2019.
  29. Der: Dynamically expandable representation for class incremental learning. In CVPR, pages 3014–3023, June 2021.
  30. Adaptive aggregation networks for class-incremental learning. In CVPR, pages 2544–2553, June 2021.
  31. Learning transferable visual models from natural language supervision. In ICML, pages 8748–8763. PMLR, 2021.
  32. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020.
  33. Learning to prompt for continual learning. In CVPR, pages 139–149, 2022.
  34. S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho, editors, Advances in Neural Information Processing Systems, 2022.
  35. Few-shot class-incremental learning via relation knowledge distillation. AAAI, 35(2):1255–1263, May 2021.
  36. Semantic-aware knowledge distillation for few-shot class-incremental learning. In CVPR, pages 2534–2543, June 2021.
  37. Uncertainty-aware distillation for semi-supervised few-shot class-incremental learning. IEEE Transactions on Neural Networks and Learning Systems, pages 1–14, 2023.
  38. Overcoming catastrophic forgetting in incremental few-shot learning by finding flat minima. NeurIPS, 34:6747–6761, 2021.
  39. Memorizing complementation network for few-shot class-incremental learning. IEEE Transactions on Image Processing, 32:937–948, 2023.
  40. On the soft-subnetwork for few-shot class incremental learning. In The Eleventh International Conference on Learning Representations, 2023.
  41. Learning a unified classifier incrementally via rebalancing. In CVPR, pages 831–839, 2019.
  42. Learning a unified classifier incrementally via rebalancing. In CVPR, June 2019.
  43. End-to-end incremental learning. In ECCV, September 2018.
  44. Imagenet large scale visual recognition challenge. IJCV, 115(3):211–252, 2015.
  45. A. Krizhevsky and G. Hinton. Learning multiple layers of features from tiny images. Handbook of Systemic Autoimmune Diseases, 1(4), 2009.
  46. The caltech-ucsd birds-200-2011 dataset. california institute of technology, 2011.
  47. Pytorch: An imperative style, high-performance deep learning library. NeurIPS, 32, 2019.
  48. Dualprompt: Complementary prompting for rehearsal-free continual learning. arXiv preprint arXiv:2204.04799, 2022.
  49. Cutmix: Regularization strategy to train strong classifiers with localizable features. In ICCV, pages 6023–6032, 2019.
  50. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412, 2017.
  51. Grad-cam: Visual explanations from deep networks via gradient-based localization. In ICCV, pages 618–626, 2017.
Citations (1)
View on Semantic Scholar

Summary

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

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

GitHub

  1. GitHub - xjtuYW/KT-RCNet (2 stars)