Feed-Forward Latent Domain Adaptation

Ondrej Bohdal, Da Li, Shell Xu Hu, Timothy Hospedales

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

We study a new highly-practical problem setting that enables resource-constrained edge devices to adapt a pretrained model to their local data distributions. Recognizing that device’s data are likely to come from multiple latent domains that include a mixture of unlabelled domain-relevant and domain-irrelevant examples, we focus on the comparatively under-studied problem of latent domain adaptation. Considering limitations of edge devices, we aim to only use a pre-trained model and adapt it in a feed-forward way, without using back-propagation and without access to the source data. Modelling these realistic constraints bring us to the novel and practically important problem setting of feedforward latent domain adaptation. Our solution is to metalearn a network capable of embedding the mixed-relevance target dataset and dynamically adapting inference for target examples using cross-attention. The resulting framework leads to consistent improvements over strong ERM baselines. We also show that our framework sometimes even improves on the upper bound of domain-supervised adaptation, where only domain-relevant instances are provided for adaptation. This suggests that human annotated domain labels may not always be optimal, and raises the possibility of doing better through automated instance selection.
Original languageEnglish
Title of host publication2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
PublisherIEEE
Pages8475-8484
Number of pages10
ISBN (Electronic)979-8-3503-1892-0
DOIs
Publication statusPublished - 9 Apr 2024
EventIEEE/CVF Winter Conference on Applications of Computer Vision, 2024
- Waikoloa, United States
Duration: 4 Jan 20248 Jan 2024
https://wacv2024.thecvf.com/

Publication series

NameProceedings (IEEE Workshop on Applications of Computer Vision. Online)
PublisherIEEE
ISSN (Print)2472-6737
ISSN (Electronic)2642-9381

Conference

ConferenceIEEE/CVF Winter Conference on Applications of Computer Vision, 2024
Abbreviated titleWACV 2024
Country/TerritoryUnited States
CityWaikoloa
Period4/01/248/01/24
Internet address

Fingerprint

Dive into the research topics of 'Feed-Forward Latent Domain Adaptation'. Together they form a unique fingerprint.

Cite this