Heterogeneous Domain Adaptation Using Manifold Alignment

By Chang Wang, Sridhar Mahadevan. 2011

457 citations recorded by the source corpus.

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

Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, background): “DA/HTL as discussed also in Section 3.3, however the main difference being that while multi-view learning [250, 239] assumes that during training multi-view examples are available, in the case of HDA [203, 237, 120, 200, 264], this assumption rarely holds (see illustration in 27). On contrary, the a”
Learning with Augmented Features for Heterogeneous Domain Adaptation (2012, methodology): “To handle more general HDA tasks, other methods have been proposed to explicitly discover a common subspace (Shi et al., 2010; Wang & Mahadevan, 2011).”
Optimal Transport for Domain Adaptation (2014, methodology): “In the first family, we find methods searching for projections discriminative in both domains, either by using dot products between the source samples and the transformed target samples [1, 7, 8], by learning projections, for which labeled samples of the target domain fall on the correct side a larg”
Comprehensive integration of single-cell data (2018, unknown): “In particular, domain adaptation aims to identify correspondences across domains to combine datasets into a shared space (Blitzer et al., 2006; Wang and Mahadevan, 2010), while transfer learning enables a model trained on a reference dataset to project information onto a query experiment (Raina et…”
Deep Visual Domain Adaptation: A Survey (2018, methodology): “[124] treated each input domain as a manifold which is represented by a Laplacian matrix, and used labels rather than correspondences to align the manifolds.”
Hybrid Heterogeneous Transfer Learning through Deep Learning (2014, background): “Specifically, one can learn two different feature mappings to transform the source domain and target domain data to a latent feature space respectively (Shi et al. 2010; Prettenhofer and Stein 2010; Wang and Mahadevan 2011; Duan, Xu, and Tsang 2012).”
Heterogeneous Domain Adaptation for Multiple Classes (2014, methodology): “Thirdly, most existing HDA methods simply adopt one-vsrest strategy to learn multiple binary classifiers independently to address the multi-class issue [8, 31, 15].”
Unsupervised Visual Domain Adaptation Using Subspace Alignment (2013, methodology): “Our method is also related to manifold alignment [16, 17, 18] whose main objective is to align two datasets from two different manifolds such that they can be projected to a common subspace.”
Joint cross-domain classification and subspace learning for unsupervised adaptation (2014, background): “Domain adaptation techniques propose to overcome these issues and make use of information coming from both source and target domains during the learning process.”
Multimodal Classification of Remote Sensing Images: A Review and Future Directions (2015, methodology): “We use the semi-supervised manifold alignment algorithm (SS–MA [129], [158]) to align the probability density functions of the images in a latent space of dimension dQB þ 2 dWV2 1⁄4 20.”
Transfer learning for speech and language processing (2015, background): “factorization [17], RBM-based latent factor learning [27] , or joint transfer optimization [28], [16], [29].”
Parallel field alignment for cross media retrieval (2013, background): “And in the same period, manifold alignment methods [10, 29, 31] are proposed and shown that they are appropriate joint models for pair matching between heterogeneous data sources.”
Heterogeneous Few-Shot Model Rectification With Semantic Mapping (2020, background): “When the source and target domains have different features, heterogeneous transfer learning [30], [31] extracts structure information or subspaces between two domains [32], [33].”
Rectify Heterogeneous Models with Semantic Mapping (2018, background): “Structure information or subspaces can be found to link two domains (Shi et al., 2010; Wang & Mahadevan, 2011), where sufficient source domain examples should be provided, even the alignment between instances across domains are required (Kulis et al., 2011).”
Instilling Social to Physical: Co-Regularized Heterogeneous Transfer Learning (2016, methodology): “The other group of models including DAMA (Wang and Mahadevan 2011), HFA (Duan et al. 2012), and SHFA (Li et al. 2014), assumes the availability of abundant labelled data and aligns different domains with labels.”
A Review of Single-Source Deep Unsupervised Visual Domain Adaptation (2020, background): “homogeneity homogeneous DA dS = dT source and target data are observed in the same space [11, 23, 36, 44–106, 113–151] heterogeneous DA dS 6= dT source and target data are observed in different spaces [107–112]”