Understanding Catastrophic Forgetting: A Three-Axis Structural Analysis in Class-Incremental Learning
Resumo
O aprendizado contínuo visa integrar novos conhecimentos sem esquecer o passado. Contudo, na Aprendizagem Incremental de Classe, os modelos são avaliados por métricas agregadas que ocultam a estrutura do esquecimento catastrófico. Por isso, propõe-se um protocolo de análise para os eixos temporal, decisório e representacional para investigar quando, como e por que ele ocorre. Foram avaliadas cinco estratégias (Naive, EWC, LwF, ER, DER++) em quatro benchmarks. Os resultados mostram que modelos com desempenho global semelhante podem apresentar diferentes mecanismos de falha, causados pela não uniformidade da manifestação do esquecimento. Conclui-se que esta avaliação multidimensional é indispensável para diagnósticos precisos.Referências
Buzzega, P., Boschini, M., Porrello, A., Abati, D., and Calderara, S. (2020). Dark experience for general continual learning: A strong, simple baseline. In Advances in Neural Information Processing Systems (NeurIPS), volume 33, pages 15920–15930.
Chaudhry, A., Dokania, P. K., Ajanthan, T., and Torr, P. H. S. (2018). Riemannian walk for incremental learning: Understanding forgetting and intransigence. In Proceedings of the European Conference on Computer Vision (ECCV), pages 556–572.
Chen, H., Wang, Y., Fan, Y., Jiang, G., and Hu, Q. (2025). Reducing class-wise confusion for incremental learning with disentangled manifolds. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10121–10130.
French, R. M. (1999). Catastrophic forgetting in connectionist networks. Trends in Cognitive Sciences, 3(4):128–135.
Kirkpatrick, J., Pascanu, R., Rabinowitz, N., Veness, J., Desjardins, G., Rusu, A. A., Milan, K., Quan, J., Ramalho, T., Grabska-Barwinska, A., Hassabis, D., Clopath, C., Kumaran, D., and Hadsell, R. (2017). Overcoming catastrophic forgetting in neural networks. Proceedings of the National Academy of Sciences, 114(13):3521–3526.
Li, Z. and Hoiem, D. (2017). Learning without forgetting. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(12):2935–2947.
Lopez-Paz, D. and Ranzato, M. (2017). Gradient episodic memory for continual learning. In Advances in Neural Information Processing Systems (NeurIPS), volume 30.
McCloskey, M. and Cohen, N. J. (1989). Catastrophic interference in connectionist networks: The sequential learning problem. Psychology of Learning and Motivation, 24:109–165.
Mermillod, M., Bugaiska, A., and BONIN, P. (2013). The stability-plasticity dilemma: investigating the continuum from catastrophic forgetting to age-limited learning effects. Frontiers in Psychology, Volume 4 - 2013.
Ramasesh, V. V., Dyer, E., and Raghu, M. (2020). Anatomy of catastrophic forgetting: Hidden representations and task geometry. arXiv preprint arXiv:2007.07400.
Rebuffi, S.-A., Kolesnikov, A., Sperl, G., and Lampert, C. H. (2017). iCaRL: Incremental classifier and representation learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2001–2010.
Rolnick, D., Ahuja, A., Schwarz, J., Lillicrap, T., and Wayne, G. (2019). Experience replay for continual learning. In Advances in Neural Information Processing Systems (NeurIPS), volume 32.
van de Ven, G. M., Soures, N., and Kudithipudi, D. (2024). Continual learning and catastrophic forgetting.
van de Ven, G. M., Tuytelaars, T., and Tolias, A. S. (2022). Three types of incremental learning. Nature Machine Intelligence, 4:1185–1197.
van der Maaten, L. and Hinton, G. (2008). Visualizing data using t-SNE. Journal of Machine Learning Research, 9:2579–2605.
Yoon, J., Yang, E., Lee, J., and Hwang, S. J. (2017). Lifelong learning with dynamically expandable networks. arXiv preprint arXiv:1708.01547.
Chaudhry, A., Dokania, P. K., Ajanthan, T., and Torr, P. H. S. (2018). Riemannian walk for incremental learning: Understanding forgetting and intransigence. In Proceedings of the European Conference on Computer Vision (ECCV), pages 556–572.
Chen, H., Wang, Y., Fan, Y., Jiang, G., and Hu, Q. (2025). Reducing class-wise confusion for incremental learning with disentangled manifolds. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10121–10130.
French, R. M. (1999). Catastrophic forgetting in connectionist networks. Trends in Cognitive Sciences, 3(4):128–135.
Kirkpatrick, J., Pascanu, R., Rabinowitz, N., Veness, J., Desjardins, G., Rusu, A. A., Milan, K., Quan, J., Ramalho, T., Grabska-Barwinska, A., Hassabis, D., Clopath, C., Kumaran, D., and Hadsell, R. (2017). Overcoming catastrophic forgetting in neural networks. Proceedings of the National Academy of Sciences, 114(13):3521–3526.
Li, Z. and Hoiem, D. (2017). Learning without forgetting. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(12):2935–2947.
Lopez-Paz, D. and Ranzato, M. (2017). Gradient episodic memory for continual learning. In Advances in Neural Information Processing Systems (NeurIPS), volume 30.
McCloskey, M. and Cohen, N. J. (1989). Catastrophic interference in connectionist networks: The sequential learning problem. Psychology of Learning and Motivation, 24:109–165.
Mermillod, M., Bugaiska, A., and BONIN, P. (2013). The stability-plasticity dilemma: investigating the continuum from catastrophic forgetting to age-limited learning effects. Frontiers in Psychology, Volume 4 - 2013.
Ramasesh, V. V., Dyer, E., and Raghu, M. (2020). Anatomy of catastrophic forgetting: Hidden representations and task geometry. arXiv preprint arXiv:2007.07400.
Rebuffi, S.-A., Kolesnikov, A., Sperl, G., and Lampert, C. H. (2017). iCaRL: Incremental classifier and representation learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2001–2010.
Rolnick, D., Ahuja, A., Schwarz, J., Lillicrap, T., and Wayne, G. (2019). Experience replay for continual learning. In Advances in Neural Information Processing Systems (NeurIPS), volume 32.
van de Ven, G. M., Soures, N., and Kudithipudi, D. (2024). Continual learning and catastrophic forgetting.
van de Ven, G. M., Tuytelaars, T., and Tolias, A. S. (2022). Three types of incremental learning. Nature Machine Intelligence, 4:1185–1197.
van der Maaten, L. and Hinton, G. (2008). Visualizing data using t-SNE. Journal of Machine Learning Research, 9:2579–2605.
Yoon, J., Yang, E., Lee, J., and Hwang, S. J. (2017). Lifelong learning with dynamically expandable networks. arXiv preprint arXiv:1708.01547.
Publicado
19/10/2026
Como Citar
SARMENTO, Gabriel F.; LEITE, Liedson D. O.; OLIVEIRA, Thyago T. M. de; CAVALCANTE, Rodolfo C.; MINKU, Leandro L.; OLIVEIRA, Gustavo H. F. de M..
Understanding Catastrophic Forgetting: A Three-Axis Structural Analysis in Class-Incremental Learning. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 2433-2444.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.28140.
