Abstract
Class incremental learning (CIL) has been attracting increasing attention in computer vision and machine learning communities, where a well-known issue is catastrophic forgetting. To mitigate this issue, a popular approach is to utilize the replay-based strategy, which stores a small portion of past data and replays it when learning new tasks. However, selecting valuable samples from previous classes for replaying remains an open problem in class incremental learning. In this paper, we propose a novel sample selection strategy aimed at maintaining effective samples from old classes to address the catastrophic forgetting issue. Specifically, we employ the influence function to evaluate the impact of each sample on model performance, and then select important samples for replay. However, given the potential redundancy among selected samples when only considering importance, we also develop a diversity strategy to select not only important but also diverse samples from old classes. We conduct extensive empirical validations on the CIFAR10 and CIFAR100 datasets and the results demonstrate that our proposed method outperforms the baselines, effectively alleviating the catastrophic forgetting issue in class incremental learning.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Li, X.: Referencing unlabelled world data to prevent catastrophic forgetting in class-incremental learning, Ph.D. thesis, Virginia Tech (2022)
Mai, Z., Li, R., Jeong, J., Quispe, D., Kim, H., Sanner, S.: Online continual learning in image classification: an empirical survey. Neurocomputing 469, 28–51 (2022)
Tang, Y.M., Peng, Y.X., Zheng, W.S.: Learning to imagine: diversify memory for incremental learning using unlabeled data. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 9549–9558 (2022)
Leo, J., Kalita, J.: Incremental deep neural network learning using classification confidence thresholding. IEEE Trans. Neural Networks Learn. Syst. 33(12), 7706–7716 (2021)
Feng, K., Li, C., Zhang, X., Zhou, J.: Towards open temporal graph neural networks. arXiv preprint arXiv:2303.15015 (2023)
Li, Z., Hoiem, D.: Learning without forgetting. IEEE Trans. Pattern Anal. Mach. Intell. 40(12), 2935–2947 (2017)
Yoon, J., Yang, E., Lee, J., Hwang, S.J.: Lifelong learning with dynamically expandable networks. arXiv preprint arXiv:1708.01547 (2017)
Aljundi, R., Chakravarty, P., Tuytelaars, T.: Expert gate: lifelong learning with a network of experts. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 3366–3375 (2017)
Robins, A.: Catastrophic forgetting, rehearsal and pseudorehearsal. Connect. Sci. 7(2), 123–146 (1995)
Shin, H., Lee, J.K., Kim, J., Kim, J.: Continual learning with deep generative replay. In: Advances in Neural Information Processing Systems, vol. 30 (2017)
Lopez-Paz, D., Ranzato, M.: Gradient episodic memory for continual learning. In: Advances in Neural Information Processing Systems, vol. 30 (2017)
Saha, G., Roy, K.: Saliency guided experience packing for replay in continual learning. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 5273–5283 (2023)
Tiwari, R., Killamsetty, K., Iyer, R., Shenoy, P.: GCR: gradient coreset based replay buffer selection for continual learning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 99–108 (2022)
Hao, Y., Fu, Y., Jiang, Y.G., Tian, Q.: An end-to-end architecture for class-incremental object detection with knowledge distillation. In: 2019 IEEE International Conference on Multimedia and Expo (ICME), pp. 1–6. IEEE (2019)
Rusu, A.A., et al.: Progressive neural networks. arXiv preprint arXiv:1606.04671 (2016)
Mallya, A., Lazebnik, S.: PackNet: adding multiple tasks to a single network by iterative pruning. In: Proceedings of the IEEE conference on Computer Vision and Pattern Recognition, pp. 7765–7773 (2018)
Saha, G., Garg, I., Ankit, A., Roy, K.: Space: structured compression and sharing of representational space for continual learning. IEEE Access 9, 150480–150494 (2021)
Serra, J., Suris, D., Miron, M., Karatzoglou, A.: Overcoming catastrophic forgetting with hard attention to the task. In: International Conference on Machine Learning, pp. 4548–4557. PMLR (2018)
Lee, S., Ha, J., Zhang, D., Kim, G.: A neural dirichlet process mixture model for task-free continual learning. arXiv preprint arXiv:2001.00689 (2020)
Chandra, D.S., Varshney, S., Srijith, P., Gupta, S.: Continual learning with dependency preserving hypernetworks. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 2339–2348 (2023)
Lee, S.W., Kim, J.H., Jun, J., Ha, J.W., Zhang, B.T.: Overcoming catastrophic forgetting by incremental moment matching. In: Advances in Neural Information Processing Systems, vol. 30 (2017)
Chaudhry, A., Ranzato, M., Rohrbach, M., Elhoseiny, M.: Efficient lifelong learning with a-gem. arXiv preprint arXiv:1812.00420 (2018)
Li, Z., Hoiem, D.: Learning without forgetting. IEEE Trans. Pattern Anal. Mach. Intell. 40(12), 2935–2947 (2017)
Feng, K., Li, C., Yuan, Y., Wang, G.: FreeKD: free-direction knowledge distillation for graph neural networks. In: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp. 357–366 (2022)
Zenke, F., Poole, B., Ganguli, S.: Continual learning through synaptic intelligence. In: International Conference on Machine Learning, pp. 3987–3995. PMLR (2017)
Dhar, P., Singh, R.V., Peng, K.C., Wu, Z., Chellappa, R.: Learning without memorizing. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5138–5146 (2019)
Nguyen, C.V., Li, Y., Bui, T.D., Turner, R.E.: Variational continual learning. arXiv preprint arXiv:1710.10628 (2017)
Aljundi, R., Caccia, L., Belilovsky, E., Caccia, M., Lin, M., Charlin, L., Tuytelaars, T.: Online continual learning with maximally interfered retrieval. CoRR abs/1908.04742 (2019), http://arxiv.org/abs/1908.04742
Riemer, M., et al.: Learning to learn without forgetting by maximizing transfer and minimizing interference. arXiv preprint arXiv:1810.11910 (2018)
Buzzega, P., Boschini, M., Porrello, A., Abati, D., Calderara, S.: Dark experience for general continual learning: a strong, simple baseline. Adv. Neural. Inf. Process. Syst. 33, 15920–15930 (2020)
Chaudhry, A., Gordo, A., Dokania, P., Torr, P., Lopez-Paz, D.: Using hindsight to anchor past knowledge in continual learning. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, pp. 6993–7001 (2021)
Ebrahimi, S., et al.: Remembering for the right reasons: explanations reduce catastrophic forgetting. Appl. AI Lett. 2(4), e44 (2021)
Chaudhry, A., et al.: Continual learning with tiny episodic memories. arXiv:1902.10486 (2019)
Rebuffi, S.A., Kolesnikov, A., Sperl, G., Lampert, C.H.: ICaRL: incremental classifier and representation learning. In: Proceedings of the IEEE conference on Computer Vision and Pattern Recognition, pp. 2001–2010 (2017)
Feng, K., Li, C., Zhang, X., Zhou, J.: Towards open temporal graph neural networks. In: The Eleventh International Conference on Learning Representations (2023). https://openreview.net/forum?id=N9Pk5iSCzAn
Lapedriza, À., Pirsiavash, H., Bylinskii, Z., Torralba, A.: Are all training examples equally valuable? CoRR abs/1311.6510 (2013), http://arxiv.org/abs/1311.6510
Koh, P.W., Liang, P.: Understanding black-box predictions via influence functions (2020)
Zhang, J., et al.: Class-incremental learning via deep model consolidation. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 1131–1140 (2020)
Wu, Y., et al.: Large scale incremental learning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 374–382 (2019)
Yoon, J., Madaan, D., Yang, E., Hwang, S.J.: Online coreset selection for rehearsal-based continual learning. arXiv preprint arXiv:2106.01085 (2021)
Castro, F.M., Marín-Jiménez, M.J., Guil, N., Schmid, C., Alahari, K.: End-to-end incremental learning. In: Proceedings of the European Conference on Computer Vision (ECCV), pp. 233–248 (2018)
Liu, Z., Ding, H., Zhong, H., Li, W., Dai, J., He, C.: Influence selection for active learning. CoRR abs/2108.09331 (2021), https://arxiv.org/abs/2108.09331
Acknowledgements
This work was supported by the NSFC under Grants 62122013, U2001211. This work was also supported by the Innovative Development Joint Fund Key Projects of Shandong NSF under Grants ZR2022LZH007.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Li, M., Yan, Z., Li, C. (2023). Class Incremental Learning with Important and Diverse Memory. In: Lu, H., et al. Image and Graphics . ICIG 2023. Lecture Notes in Computer Science, vol 14358. Springer, Cham. https://doi.org/10.1007/978-3-031-46314-3_13
Download citation
DOI: https://doi.org/10.1007/978-3-031-46314-3_13
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-46313-6
Online ISBN: 978-3-031-46314-3
eBook Packages: Computer ScienceComputer Science (R0)