Unsupervised Cross-domain Image Classification by Distance Metric Guided Feature Alignment
Meng Q, Rueckert D, Kainz B (2020)
Publication Type: Conference contribution
Publication year: 2020
Journal
Publisher: Springer Science and Business Media Deutschland GmbH
Book Volume: 12437 LNCS
Pages Range: 146-157
Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Event location: Lima, PER
ISBN: 9783030603335
DOI: 10.1007/978-3-030-60334-2_15
Abstract
Learning deep neural networks that are generalizable across different domains remains a challenge due to the problem of domain shift. Unsupervised domain adaptation is a promising avenue which transfers knowledge from a source domain to a target domain without using any labels in the target domain. Contemporary techniques focus on extracting domain-invariant features using domain adversarial training. However, these techniques neglect to learn discriminative class boundaries in the latent representation space on a target domain and yield limited adaptation performance. To address this problem, we propose distance metric guided feature alignment (MetFA) to extract discriminative as well as domain-invariant features on both source and target domains. The proposed MetFA method explicitly and directly learns the latent representation without using domain adversarial training. Our model integrates class distribution alignment to transfer semantic knowledge from a source domain to a target domain. We evaluate the proposed method on fetal ultrasound datasets for cross-device image classification. Experimental results demonstrate that the proposed method outperforms the state-of-the-art and enables model generalization.
Authors with CRIS profile
Involved external institutions
United Kingdom (GB)How to cite
APA:
Meng, Q., Rueckert, D., & Kainz, B. (2020). Unsupervised Cross-domain Image Classification by Distance Metric Guided Feature Alignment. In Yipeng Hu, Roxane Licandro, J. Alison Noble, Jana Hutter, Andrew Melbourne, Stephen Aylward, Esra Abaci Turk, Jordina Torrents Barrena, Jordina Torrents Barrena (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 146-157). Lima, PER: Springer Science and Business Media Deutschland GmbH.
MLA:
Meng, Qingjie, Daniel Rueckert, and Bernhard Kainz. "Unsupervised Cross-domain Image Classification by Distance Metric Guided Feature Alignment." Proceedings of the 1st International Workshop on Advances in Simplifying Medical UltraSound, ASMUS 2020, and the 5th International Workshop on Perinatal, Preterm and Paediatric Image Analysis, PIPPI 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020, Lima, PER Ed. Yipeng Hu, Roxane Licandro, J. Alison Noble, Jana Hutter, Andrew Melbourne, Stephen Aylward, Esra Abaci Turk, Jordina Torrents Barrena, Jordina Torrents Barrena, Springer Science and Business Media Deutschland GmbH, 2020. 146-157.
BibTeX: Download