Robust visual domain adaptation with low-rank reconstruction

By I-Hong Jhuo, Dong Liu, D. T. Lee, Shih-Fu Chang. 2012

344 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Mechanics, Epidemiology

How researchers see this paper

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

In Artificial Intelligence

In Computational Mechanics

In Epidemiology

Citations in the corpus

Taskonomy: Disentangling Task Transfer Learning (2018, neutral): “webcam images to D-SLR [45], while the task is kept the same.”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, background): “e preserving the underlying structure of each domain. This is done by representing each domains with a Laplacian matrix that captures the closeness of the instances sharing the same label. The RDALRR [120] mentioned in Section 3.2 (see also Figure 11) transforms each source domain into an intermedia”
Predicting Ground-Level Scene Layout from Aerial Imagery (2016, background): “[13] propose a low-rank reconstruction approach where the source features are transformed to an intermediate representation in which they can be linearly reconstructed by the target samples.”
DeepJDOT: Deep Joint distribution optimal transport for unsupervised domain adaptation (2018, background): “Earlier approaches consider projections aligning data spaces to each other [2, 11, 12], thus trying to exploit shift-invariant information to match the domains in their original (input) space.”
Joint cross-domain classification and subspace learning for unsupervised adaptation (2014, methodology): “Finally, in previous work the performance of the adaptive methods have been often evaluated by tuning the model parameters on the target data [12, 24] or by fixing them to default values [21].”
Structure-Preserved Multi-source Domain Adaptation (2016, background): “the adaptation process to shift the discriminative function from the source domain and applies it for the target domain [35]; Low-rank Transfer Subspace Learning (LTSL) first learns a subspace with the low-rank constraints, then applies PCA or LDA for the adaptation [29]; SGF trains the discriminati”
Cross-Domain Object Recognition Via Input-Output Kernel Analysis (2013, methodology): “Source Target NC A-SVM ITML [22] symm [3] ARC-t [4] RDALR [11] DA-IOKL”
Aligning Domain-Specific Distribution and Classifier for Cross-Domain Classification from Multiple Sources (2019, methodology): “The research originates from A-SVM (Yang, Yan, and Hauptmann 2007) that leverages the ensemble of source-specific classifiers to tune the target categorization model, and there have been a variety of shallow models invented to tackle the MUDA problem (Duan, Xu, and Tsang 2012; Jhuo et al. 2012; Liu,”
Generalized Transfer Subspace Learning Through Low-Rank Constraint (2014, methodology): “Finally, robust visual domain adaptation with low-rank construction (RDALR) (Jhuo et al. 2012) is very similar to our proposed framework, however, it uses a naive linear projection to cou- ple source and target domains.”
Transfer Joint Matching for Unsupervised Domain Adaptation (2014, background): “Domain adaptation has proven to be promising in image classiﬁcation [25, 13] and tagging [23, 26], object recognition [15, 2, 7, 10], and feature learning [14, 12, 22] In cross-domain problems, the source and target data are usually sampled from different probability distributions.”
Deep Domain Generalization With Structured Low-Rank Constraint (2018, background): “Multi-source adaptation problem [4], [12] is similar to domain generalization, since they both manage to extract the source knowledge and handle the distribution difference across sources.”
Generalized Domain-Adaptive Dictionaries (2013, methodology): “[11] learnt a transformation of source data onto target space, such that the joint representation is low-rank.”
Subspace Interpolation via Dictionary Learning for Unsupervised Domain Adaptation (2013, background): “This is often known as the domain adaptation problem which has recently drawn much attention in the computer vision community [28, 14, 13, 17].”
Harnessing Lab Knowledge for Real-World Action Recognition (2014, methodology): “Low-rank and group sparsity structure are exploited in Jhuo et al. (2012) to reduce the distribution disparity between two domains, by which the data in the source domain are transformed into an intermediate representation that can be linearly reconstructed by the data of the target domain.”
Learning Sparse FRAME Models for Natural Image Patterns (2014, methodology): “main datasets, and compare with published results [49] [24] [23] [61] [30] [28] [50].”