Model-induced Generalization Error Bound for Information-theoretic Representation Learning in Source-data-free Unsupervised Domain Adaptation

Many unsupervised domain adaptation (UDA) methods have been developed and have achieved promising results in various pattern recognition tasks. However, most existing methods assume that raw source data are available in the target domain when transferring knowledge from the source to the target doma...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on image processing 2022-01, Vol.31, p.1-1
Main Authors: Yang, Baoyao, Yeh, Hao-Wei, Harada, Tatsuya, Yuen, Pong C.
Format: Article
Language:English
Subjects:
Ablation
Adaptation
Adaptation models
Computational modeling
Data models
Data privacy
Domains
Errors
Information theory
Optimization
Pattern recognition
Predictive models
Representations
Theorems
Upper bound
Upper bounds
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Many unsupervised domain adaptation (UDA) methods have been developed and have achieved promising results in various pattern recognition tasks. However, most existing methods assume that raw source data are available in the target domain when transferring knowledge from the source to the target domain. Due to the emerging regulations on data privacy, the availability of source data cannot be guaranteed when applying UDA methods in a new domain. The lack of source data makes UDA more challenging, and most existing methods are no longer applicable. To handle this issue, this paper analyzes the cross-domain representations in source-data-free unsupervised domain adaptation (SF-UDA). A new theorem is derived to bound the target-domain prediction error using the trained source model instead of the source data. On the basis of the proposed theorem, information bottleneck theory is introduced to minimize the generalization upper bound of the target-domain prediction error, thereby achieving domain adaptation. The minimization is implemented in a variational inference framework using a newly developed latent alignment variational autoencoder (LA-VAE). The experimental results show good performance of the proposed method in several cross-dataset classification tasks without using source data. Ablation studies and feature visualization also validate the effectiveness of our method in SF-UDA.
ISSN:1057-7149
1941-0042
DOI:10.1109/TIP.2021.3130530