Domain adaptationoptimal transportlabel propagationDomain adaptation is a subfield of statistical learning theory that takes into account the shift between the distribution of training and test data, typically known as source and target domains, respectively. In this context, this paper presents an ...
In this paper, we propose a novel approach for unsupervised domain adaptation, that relates notions of optimal transport, learning probability measures and unsupervised learning. The proposed approach, HOT-DA, is based on a hierarchical formulation of optimal transport, that leverages beyond the ...
Domain adaptation from one data space (or domain) to another is one of the most challenging tasks of modern data analytics. If the adaptation is done correctly, models built on a specific data space become more robust when confronted to data depicting th
这个优化也是两步交替优化, 寻找最优的迁移, 同时去做下游任务, 后边有篇 DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation, 就把它放在深度学习上了.ImageA Bound on the Target Error 记errT(f)= def E(x,y)∼PtL(y,f(x)) ...
Unif i ed Optimal Transport Framework for UniversalDomain AdaptationWanxing Chang 1 Ye Shi 1,3∗ Hoang Duong Tuan 2 Jingya Wang 1,3∗1 ShanghaiTech University 2 University of Technology Sydney3 Shanghai Engineering Research Center of Intelligent Vision and Imaging{changwx,shiye,wangjingya}@shan...
In this paper, we propose a novel approach for unsupervised domain adaptation that relates notions of optimal transport, learning probability measures, and unsupervised learning. The proposed approach, HOT-DA, is based on a hierarchical formulation of optimal transport that leverages beyond the geometric...
Optimal Transport for Multi-source Domain Adaptation under Target Shift 这篇论文在 Optical Transport 解决域适应的基础上,提出了域间类别不平衡的问题并加以解决。 (a)图是未做 Optical Transport 之前,…
Recently, extensive researches have been proposed to address the UDA problem, which aims to learn transferrable models for the unlabeled target domain. Among them, the optimal transport is a promising metric to align the representations of the source and target domains. However, most exist...
3Optimal Transport for Domain Adaptation Our proposal is based on optimal transport. After recalling the associated basic notions and its relation with domain adaptation, we detail the JDOT method [5], which is the starting point of our proposition. ...
[9,5].Those divergences usually require that the distributions share a common support to be de,ned.A particular case is found in the use of optimal transport, introduced for domain adaptation by[13, 14].T is then de,ned to be a push-forward operator such that Ps(X)=Pt(T(X))and ...