Visual Domain Adaptation with Manifold Embedded Distribution Alignment

By Jindong Wang, Wenjie Feng, Yiqiang Chen, Han Yu, Meiyu Huang, Philip S. Yu. ACM Multimedia, 2018

651 citations recorded by the source corpus.

Visual domain adaptation aims to learn robust classifiers for the target domain by leveraging knowledge from a source domain. Existing methods either attempt to align the cross-domain distributions, or perform manifold subspace learning. However, there are two significant challenges: (1) degenerated feature transformation, which means that distribution alignment is often performed in the original feature space, where feature distortions are hard to overcome. On the other hand, subspace learning is not sufficient to reduce the distribution divergence. (2) unevaluated distribution alignment, which means that existing distribution alignment methods only align the marginal and conditional distributions with equal importance, while they fail to evaluate the different importance of these two distributions in real applications. In this paper, we propose a Manifold Embedded Distribution Alignment (MEDA) approach to address these challenges. MEDA learns a domain-invariant classifier in Grassmann manifold with structural risk minimization, while performing dynamic distribution alignment to quantitatively account for the relative importance of marginal and conditional distributions. To the best of our knowledge, MEDA is the first attempt to perform dynamic distribution alignment for manifold domain adaptation. Extensive experiments demonstrate that MEDA shows significant improvements in classification accuracy compared to state-of-the-art traditional and deep methods.

Subfields: Artificial Intelligence, Cancer Research, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/cc7c62516fbbf5d872d874bd598c8ab8ba903808

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 Cancer Research

In Computer Vision and Pattern Recognition

Citations in the corpus

FedHealth: A Federated Transfer Learning Framework for Wearable Healthcare (2019, methodology): “We compare its performance with two methods: 1) ﬁne-tuning, which only ﬁne-tunes the network on each subject without explicitly reducing the distribution divergence between domains; and 2) transfer with MMD (Maximum Mean Discrepancy) [ Wang et al. , 2018b ] , which replaces the alignment loss with M”
Reliable Weighted Optimal Transport for Unsupervised Domain Adaptation (2020, background): “However, existing optimal transport approaches in domain adaptation do not consider intra-domain structure [44] of both domains, only achieving coarse pair-wise matching.”
Deep Subdomain Adaptation Network for Image Classification (2020, methodology): “CMMD [7], [23], [33] is a special case of”
Towards Out-Of-Distribution Generalization: A Survey (2021, background): “Some works minimize the divergence of feature distributions by minimizing the maximum mean discrepancy (MMD) distance [37], [39], [50], [51], Wasserstein distance [52], the second order correlation [53], [54], [55], etc., for DA or DG. Feature normalization.”
Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift (2022, background): “Domain adaptation (Tzeng et al., 2017; Ganin et al., 2016; Wang et al., 2018) and domain generalization (Wang et al.”
Contrastive Adaptation Network for Unsupervised Domain Adaptation (2019, methodology): “our method performs class-aware domain alignment. Discriminativedomain-invariantfeaturelearning. Some previous works pay efforts to learn more disciminative features while performing domain alignment [35,13,31,32, 28,39]. Adversarial Dropout Regularization (ADR) [31] and Maximum Classiﬁer Discrepanc”
Generalizing to Unseen Domains: A Survey on Domain Generalization (2021, methodology): “it is more feasible to perform dynamic distribution matching [189], [190].”
Transfer Learning with Dynamic Adversarial Adaptation Network (2019, methodology): “• Deep residual learning [1] • Deep Domain Confusion (DDC) [33] • Deep Adaptation Network (DAN) [32] • Residual Transfer Network (RTN) [41] • Domain Adversarial Neural Networks (DANN) [18] • Deep CORAL (D-CORAL) [17] • Joint Adaptation Networks (JAN) [21] • Multi-Adversarial Domain Adaptation (MADA)”
AdaRNN: Adaptive Learning and Forecasting of Time Series (2021, methodology): “DA often performs instance re-weighting or feature transfer to reduce the distribution divergence in training and test data [12, 44, 47, 48, 56].”
Moment Matching for Multi-Source Domain Adaptation (2018, methodology): “Inspired by the kernel two-sample test [12], Maximum Mean Discrepancy (MMD) is applied to reduce distribution shift in various methods [27, 41, 9, 44].”