Unsupervised Domain Adaptation with a Relaxed Covariate Shift Assumption

By Tameem Adel, Han Zhao, Alexander Wong. 2017

28 citations recorded by the source corpus.

Domain adaptation addresses learning tasks where training is performed on data from one domain whereas testing is performed on data belonging to a different but related domain. Assumptions about the relationship between the source and target domains should lead to tractable solutions on the one hand, and be realistic on the other hand. Here we propose a generative domain adaptation model that allows for modelling different assumptions about this relationship, among which is a newly introduced assumption that replaces covariate shift with a possibly more realistic assumption without losing tractability due to the efficient variational inference procedure developed. In addition to the ability to model less restrictive relationships between source and target, modelling can be performed without any target labeled data (unsupervised domain adaptation). We also provide a Rademacher complexity bound of the proposed algorithm. We evaluate the model on the Amazon reviews and the CVC pedestrian detection datasets.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Continual Learning with Adaptive Weights (CLAW) (2019, methodology): “Our model is also related to the multi-task learning approach (Caruana, 1997; Heskes, 2000; Bakker & Heskes, 2003; Adel et al., 2017; Zhao et al., 2019a; Stickland & Murray, 2019; Zhao et al., 2019b).”
Efficient Multitask Feature and Relationship Learning (2017, background): “…et al., 2017, Argyriou et al., 2007, 2008, Caruana, 1997, Chen et al., 2011, 2012, Evgeniou and Pontil, 2004, Jacob et al., 2009, Jawanpuria et al., 2015, Kato et al., 2008, Li et al., 2015, Liu et al., 2009, Zhang and Schneider, 2010, Zhang and Yeung, 2010a, Zhao et al., 2017, 2018, 2019a,b).”
Self-Ensembling Attention Networks: Addressing Domain Shift for Semantic Segmentation (2019, methodology): “However, the performance of these models on real images is usually not satisfactory due to the phenomenon of domain shift (Adel, Zhao, and Wong 2017), which is caused by factors like imaging devices (graphics engines vs. RGB sensors), illumination, distortion, and shadows in different domains,…”
A Review of Single-Source Deep Unsupervised Visual Domain Adaptation (2020, background): “While most of the work we discussed so far focuses on learning domain-invariant representations, methods based on estimating the importance ratio of density functions between target and source domains are also abundant in the literature [28, 186–189].”
On Learning Invariant Representations for Domain Adaptation (2019, background): “, 2009a; Mansour & Schain, 2012) and algorithms (Glorot et al., 2011; Becker et al., 2013; Ajakan et al., 2014; Adel et al., 2017; Pei et al., 2018) under this setting are abundant.”