EUVIMEDEUVIMED entdecken
Crossref

Continual Learning: Forget-Free Winning Subnetworks for Video Representations

Haeyong Kang, Jaehong Yoon, Sung Ju Hwang, Chang D Yoo

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2025-03 · 47 (3) · 2140-2156

DOI: 10.1109/tpami.2024.3518588 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025-03. Verfasst von Haeyong Kang und 3 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 120 Literaturangaben. Gefördert von: Institute for Information and Communications Technology Promotion.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Inspired by the Lottery Ticket Hypothesis (LTH), which highlights the existence of efficient subnetworks within larger, dense networks, a high-performing Winning Subnetwork (WSN) in terms of task performance under appropriate sparsity conditions is considered for various continual learning tasks. It leverages pre-existing weights from dense networks to achieve efficient learning in Task Incremental Learning (TIL) and Task-agnostic Incremental Learning (TaIL) scenarios. In Few-Shot Class Incremental Learning (FSCIL), a variation of WSN referred to as the Soft subnetwork (SoftNet) is designed to prevent overfitting when the data samples are scarce. Furthermore, the sparse reuse of WSN weights is considered for Video Incremental Learning (VIL). The use of Fourier Subneural Operator (FSO) within WSN is considered. It enables compact encoding of videos and identifies reusable subnetworks across varying bandwidths. We have integrated FSO into different architectural frameworks for continual learning, including VIL, TIL, and FSCIL. Our comprehensive experiments demonstrate FSO's effectiveness, significantly improving task performance at various convolutional representational levels. Specifically, FSO enhances higher-layer performance in TIL and FSCIL and lower-layer performance in VIL.

Förderung

  • Institute for Information and Communications Technology Promotion

Zitieren

APA

Kang, H., Yoon, J., Hwang, S. J., & Yoo, C. D. (2025). Continual Learning: Forget-Free Winning Subnetworks for Video Representations. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(3), 2140-2156. https://doi.org/10.1109/tpami.2024.3518588

Vancouver

Kang H, Yoon J, Hwang SJ, Yoo CD. Continual Learning: Forget-Free Winning Subnetworks for Video Representations. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2025;47(3):2140-2156. doi:10.1109/tpami.2024.3518588

BibTeX
@article{Kang2025,
  author = {Kang, Haeyong and Yoon, Jaehong and Hwang, Sung Ju and Yoo, Chang D},
  title = {Continual Learning: Forget-Free Winning Subnetworks for Video Representations},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2025},
  volume = {47},
  number = {3},
  pages = {2140--2156},
  doi = {10.1109/tpami.2024.3518588}
}
Kostenlos starten →