Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding
Proceedings of the AAAI Conference on Artificial Intelligence · 24.03.2024 · 38 (12) · 13464-13473
DOI: 10.1609/aaai.v38i12.29249 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 24.03.2024. Verfasst von Depeng Li und 5 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref).
Abstract (Originalsprache)
Deep neural networks are susceptible to catastrophic forgetting when trained on sequential tasks. Various continual learning (CL) methods often rely on exemplar buffers or/and network expansion for balancing model stability and plasticity, which, however, compromises their practical value due to privacy and memory concerns. Instead, this paper considers a strict yet realistic setting, where the training data from previous tasks is unavailable and the model size remains relatively constant during sequential training. To achieve such desiderata, we propose a conceptually simple yet effective method that attributes forgetting to layer-wise parameter overwriting and the resulting decision boundary distortion. This is achieved by the synergy between two key components: HSIC-Bottleneck Orthogonalization (HBO) implements non-overwritten parameter updates mediated by Hilbert-Schmidt independence criterion in an orthogonal space and EquiAngular Embedding (EAE) enhances decision boundary adaptation between old and new tasks with predefined basis vectors. Extensive experiments demonstrate that our method achieves competitive accuracy performance, even with absolute superiority of zero exemplar buffer and 1.02x the base model.
Zitieren
APA
Li, D., Wang, T., Chen, J., Ren, Q., Kawaguchi, K., & Zeng, Z. (2024). Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding. Proceedings of the AAAI Conference on Artificial Intelligence, 38(12), 13464-13473. https://doi.org/10.1609/aaai.v38i12.29249
Vancouver
Li D, Wang T, Chen J, Ren Q, Kawaguchi K, Zeng Z. Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding. Proceedings of the AAAI Conference on Artificial Intelligence. 2024;38(12):13464-13473. doi:10.1609/aaai.v38i12.29249
BibTeX
@article{Li2024,
author = {Li, Depeng and Wang, Tianqi and Chen, Junwei and Ren, Qining and Kawaguchi, Kenji and Zeng, Zhigang},
title = {Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2024},
volume = {38},
number = {12},
pages = {13464--13473},
doi = {10.1609/aaai.v38i12.29249}
}
EUVIMED