Abstract:Partial domain adaptation (PDA) attempts to learn transferable models from a large-scale labeled source domain to a small unlabeled target domain with fewer classes, which has attracted a recent surge of interest in transfer learning. Most conventional PDA approaches endeavor to design delicate source weighting schemes by leveraging target predictions to align cross-domain distributions in the shared class space. Accordingly, two crucial issues are overlooked in these methods. First, target prediction is a double-edged sword, and inaccurate predictions will result in negative transfer inevitably. Second, not all target samples have equal transferability during the adaptation; thus, "ambiguous" target data predicted with high uncertainty should be paid more attentions. In this article, we propose a critical classes and samples discovering network (CSDN) to identify the most relevant source classes and critical target samples, such that more precise cross-domain alignment in the shared label space could be enforced by co-training two diverse classifiers. Specifically, during the training process, CSDN introduces an adaptive source class weighting scheme to select the most relevant classes dynamically. Meanwhile, based on the designed target ambiguous score, CSDN emphasizes more on ambiguous target samples with larger inconsistent predictions to enable fine-grained alignment. Taking a step further, the weighting schemes in CSDN can be easily coupled with other PDA and DA methods to further boost their performance, thereby demonstrating its flexibility. Extensive experiments verify that CSDN attains excellent results compared to state of the arts on four highly competitive benchmark datasets.

Selective Adversarial Adaptation Learning Via Exclusive Regularization for Partial Domain Adaptation

Selective Transfer with Reinforced Transfer Network for Partial Domain Adaptation.

Federated Domain Adaptation via Pseudo-label Refinement

Partial Adversarial Domain Adaptation

Self-Adaptive Partial Domain Adaptation

Partial Domain Adaptation for Relation Extraction Based on Adversarial Learning

Discriminative Partial Domain Adversarial Network

Multi-Weight Partial Domain Adaptation.

Adversarial Reweighting for Partial Domain Adaptation.

Partial Domain Adaptation Based on Shared Class Oriented Adversarial Network.

Domain Adversarial Reinforcement Learning for Partial Domain Adaptation

Attention-based Adversarial Partial Domain Adaptation

Self-adaptive Re-weighted Adversarial Domain Adaptation

Adversarial Partial Domain Adaptation by Cycle Inconsistency.

Partial Transfer Learning with Selective Adversarial Networks

Class Conditional Alignment for Partial Domain Adaptation

Critical Classes and Samples Discovering for Partial Domain Adaptation

Learning Target Domain Specific Classifier for Partial Domain Adaptation

Learning to Transfer Examples for Partial Domain Adaptation

From Big to Small: Adaptive Learning to Partial-Set Domains

Adversarial Reweighting with -Power Maximization for Domain Adaptation