Abstract:Adversarial learning has proven to be an effective method for capturing transferable features for unsupervised domain adaptation. However, some existing conditional adversarial domain adaptation methods assign equal importance to different samples, ignoring the fact that hard-to-transfer samples might damage the conditional adversarial adaptation procedure. Meanwhile, some methods can only roughly align marginal distributions across domains, but cannot ensure category distributions alignment, causing classifiers to make uncertain or even wrong predictions for some target data. Furthermore, we find that the feature norms of real images usually follow a complex distribution, so directly matching the mean feature norms of two domains cannot effectively reduce the statistical discrepancy of feature norms and may potentially induce feature degradation. In this paper, we develop a Trust-aware Conditional Adversarial Domain Adaptation (TCADA) method for solving the aforementioned issues. To quantify data transferability, we suggest utilizing posterior probability modeled by a Gaussian-uniform mixture, which effectively facilitates conditional domain alignment. Based on this posterior probability, a confidence-guided alignment strategy is presented to promote precise alignment of category distributions and accelerate the learning of shared features. Moreover, a novel optimal transport-based strategy is introduced to align the feature norms and facilitate shared features becoming more informative. To encourage classifiers to make more accurate predictions for target data, we also design a mixed information-guided entropy regularization term to promote deep features being away from the decision boundaries. Extensive experiments show that our method greatly improves transfer performance on various tasks.

Low-Rank Optimal Transport for Robust Domain Adaptation

Uncertainty-guided Joint Unbalanced Optimal Transport for Unsupervised Domain Adaptation

Trust-aware Conditional Adversarial Domain Adaptation with Feature Norm Alignment.

Reliable Weighted Optimal Transport For Unsupervised Domain Adaptation

HOMDA: High-Order Moment-Based Domain Alignment for Unsupervised Domain Adaptation

Joint Distribution Optimal Transportation for Domain Adaptation

Semi-Supervised Optimal Transport for Heterogeneous Domain Adaptation

Decomposed-distance weighted optimal transport for unsupervised domain adaptation

Finding Robust Transfer Features for Unsupervised Domain Adaptation.

Metric Learning in Optimal Transport for Domain Adaptation

Domain Adaptation As Optimal Transport on Grassmann Manifolds

Incremental Confidence Sampling with Optimal Transport for Domain Adaptation

Few-Shot Domain Adaptation via Mixup Optimal Transport

Optimal Transport for Domain Adaptation through Gaussian Mixture Models

MOT: Masked Optimal Transport for Partial Domain Adaptation

Label-Noise Robust Domain Adaptation

Probability-Polarized Optimal Transport for Unsupervised Domain Adaptation

Neural domain alignment for spoken language recognition based on optimal transport

Joint Partial Optimal Transport for Open Set Domain Adaptation

Prototypical Partial Optimal Transport for Universal Domain Adaptation

A Risk Minimization Framework for Domain Adaptation