Optimal Transport for Multi-source Domain Adaptation under Target Shift

Ievgen Redko 1 Nicolas Courty 2 Rémi Flamary 3 Devis Tuia 4
1 lahC
LHC - Laboratoire Hubert Curien [Saint Etienne]
2 OBELIX - Environment observation with complex imagery
UBS - Université de Bretagne Sud, IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : In this paper, we tackle the problem of reducing discrepancies between multiple domains, i.e. multi-source domain adaptation, and consider it under the target shift assumption: in all domains we aim to solve a classification problem with the same output classes, but with different labels proportions. This problem , generally ignored in the vast majority of domain adaptation papers, is nevertheless critical in real-world applications, and we theoretically show its impact on the success of the adaptation. Our proposed method is based on optimal transport, a theory that has been successfully used to tackle adaptation problems in machine learning. The introduced approach , Joint Class Proportion and Optimal Transport (JCPOT), performs multi-source adaptation and target shift correction simultaneously by learning the class probabilities of the unlabeled target sample and the coupling allowing to align two (or more) probability distributions. Experiments on both synthetic and real-world data (satellite image pixel classification) task show the superiority of the proposed method over the state-of-the-art.
Document type :
Conference papers
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-02082874
Contributor : Ievgen Redko <>
Submitted on : Thursday, March 28, 2019 - 4:38:17 PM
Last modification on : Wednesday, April 3, 2019 - 1:52:05 AM

File

AISTATS_jcpot.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02082874, version 1

Citation

Ievgen Redko, Nicolas Courty, Rémi Flamary, Devis Tuia. Optimal Transport for Multi-source Domain Adaptation under Target Shift. 22nd International Conference on Artificial Intelligence and Statistics (AISTATS) 2019, Apr 2019, Naha, Japan. ⟨hal-02082874⟩

Share

Metrics

Record views

70

Files downloads

34