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.

Domain adaptation image classification based on target local-neighbor geometrical information

Domain adaptation of image classification exploiting target adaptive collaborative local-neighbor representation

Domain Adaptation of Image Classification Based on Collective Target Nearest-Neighbor Representation

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

Self-Supervised Global-Local Structure Modeling for Point Cloud Domain Adaptation with Reliable Voted Pseudo Labels

Domain Adaptation on Point Clouds Via Geometry-Aware Implicits

Discriminative and Geometry-Aware Unsupervised Domain Adaptation

An Unsupervised Domain Adaptation Method for Multi-Modal Remote Sensing Image Classification

Graph-structure Constraint and Schatten P-Norm-based Unsupervised Domain Adaptation for Image Classification

Deep Domain Similarity Adaptation Networks for Across Domain Classification

Domain Adaptation Object Recognition

Unsupervised domain adaption dictionary learning for visual recognition

Synthetic Data with Global-Local Level for Adversarial-based Domain Adaptation

An Embarrassingly Simple Approach to Visual Domain Adaptation.

A New Method of Image Classification Based on Domain Adaptation

Class Distribution Alignment for Adversarial Domain Adaptation

Domain Adaptation for Remote Sensing Image Semantic Segmentation: An Integrated Approach of Contrastive Learning and Adversarial Learning

Class-Aware Domain Adaptation for Coastal Land Cover Mapping Using Optical Remote Sensing Imagery

Unsupervised Urban Scene Segmentation Via Domain Adaptation

Domain Adaption Object Detection with Global - Local Contrastive Learning and Co-training Network.

Unsupervised domain adaptation based on cluster matching and Fisher criterion for image classification