PIVOT: Prompting for Video Continual Learning

Simple item page
dc.contributor.authorVilla Ojeda, Andres Felipe
dc.contributor.authorAlcazar, Juan Leon
dc.contributor.authorAlfarra, Motasem
dc.contributor.authorAlhamoud, Kumail
dc.contributor.authorHurtado, Julio
dc.contributor.authorHeilbron, Fabian Caba
dc.contributor.authorSoto, Alvaro
dc.contributor.authorGhanem, Bernard
dc.date.accessioned2024-03-06T15:37:11Z
dc.date.available2024-03-06T15:37:11Z
dc.date.issued2023
dc.description.abstractModern machine learning pipelines are limited due to data availability, storage quotas, privacy regulations, and expensive annotation processes. These constraints make it difficult or impossible to train and update large-scale models on such dynamic annotated sets. Continual learning directly approaches this problem, with the ultimate goal of devising methods where a deep neural network effectively learns relevant patterns for new (unseen) classes, without significantly altering its performance on previously learned ones. In this paper, we address the problem of continual learning for video data. We introduce PIVOT, a novel method that leverages extensive knowledge in pre-trained models from the image domain, thereby reducing the number of trainable parameters and the associated forgetting. Unlike previous methods, ours is the first approach that effectively uses prompting mechanisms for continual learning without any in-domain pre-training. Our experiments show that PIVOT improves state-of-the-art methods by a significant 27% on the 20-task ActivityNet setup.
dc.description.funderANID - Millennium Science Initiative Program / Millennium Institute for Research on Depression and Personality-MIDAP
dc.format.extent18 páginas
dc.fuente.origenWOS
dc.identifier.doi10.1109/CVPR52729.2023.02319
dc.identifier.eisbn979-8-3503-0129-8
dc.identifier.eissn1579-3699
dc.identifier.issn1063-6919
dc.identifier.pubmedidMEDLINE:34927594
dc.identifier.urihttps://doi.org/10.1109/CVPR52729.2023.02319
dc.identifier.urihttps://repositorio.uc.cl/handle/11534/84267
dc.identifier.wosidWOS:001062531308053
dc.information.autorucEscuela de Ingeniería; Villa Ojeda, Andres Felipe; S/I; 1092267
dc.issue.numero3
dc.language.isoen
dc.nota.accesoContenido parcial
dc.pagina.final24223
dc.pagina.inicio24214
dc.relation.ispartofIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2023, Vancouver, Canadá)
dc.revistaIEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
dc.rightsacceso restringido
dc.subjectdepression
dc.subjectadolescence
dc.subjectprevention
dc.subjectearly intervention
dc.subjectInternet-based interventions
dc.subjectonline program
dc.subjectE-Health
dc.subject.ods03 Good Health and Well-being
dc.subject.odspa03 Salud y bienestar
dc.titlePIVOT: Prompting for Video Continual Learning
dc.typecomunicación de congreso
dc.volumen15
sipa.codpersvinculados1092267
sipa.indexWOS
sipa.trazabilidadCarga WOS-SCOPUS;06-03-2024
Files
Collections
Artículos de conferencia