Continual Learning: Forget-free Winning Subnetworks for Video Representations

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Kang, Haeyong, Yoon, Jaehong, Hwang, Sung Ju, Yoo, Chang D.
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866916531552649216
author Kang, Haeyong
Yoon, Jaehong
Hwang, Sung Ju
Yoo, Chang D.
author_facet Kang, Haeyong
Yoon, Jaehong
Hwang, Sung Ju
Yoo, Chang D.
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2312_11973
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Continual Learning: Forget-free Winning Subnetworks for Video Representations
Kang, Haeyong
Yoon, Jaehong
Hwang, Sung Ju
Yoo, Chang D.
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
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.
title Continual Learning: Forget-free Winning Subnetworks for Video Representations
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2312.11973