Abstract:Domain adaptation aims to exploit domain-invariant features by aligning the cross-domain distributions in the manifold subspace for applying the classifier trained on the source domain to the target domain. However, two limitations may still deteriorate their performances: (1) the influences of noisy or irrelevant features in the original feature space are ignored, which may unexpectedly hurt the classification of target samples; (2) the graph constructed directly in the original data space cannot accurately capture the inherent local manifold structures of high-dimensional data due to the curse of dimensionality, which may seriously mislead the transferable features learning. In this paper, we propose a novel approach to address these problems, referred to as joint Adaptive Dual Graph and Feature Selection for domain adaptation (ADGFS). Specifically, feature selection can characterize the relative importance of different features through a scaling factor, which enables ADGFS to not only reduce the impacts of noisy or irrelevant features on knowledge transfer but also learn informative domain-invariant features. Meanwhile, ADGFS adaptively optimizes the dual graph by learning the similarity matrices of both instance-level and feature-level graphs in the projected low-dimensional manifold subspace rather than the original high-dimensional space, such that the intrinsic local manifold structures of data can be captured precisely. Moreover, ADGFS simultaneously aligns the marginal and conditional probability distributions in the nonnegative matrix factorization framework to narrow the distribution discrepancies between the two different domains, which can adequately transfer knowledge from the source domain to the target domain. Comprehensive experiments on four benchmark datasets can demonstrate that the effectiveness of the proposed approach in cross-domain image classification.

Integration of Global and Local Metrics for Domain Adaptation Learning Via Dimensionality Reduction

Joint Domain Alignment and Discriminative Feature Learning for Unsupervised Deep Domain Adaptation

Domain Adaptation Based on the Measure of Kernel-Product Maximum Mean Discrepancy.

DAML: domain adaptation metric learning.

Kernelized Bures metric: A framework for effective domain adaptation in sensor data analysis

Adaptive Local Neighbors for Transfer Discriminative Feature Learning.

Local Weight Coupled Network: Multi-Modal Unequal Semi-Supervised Domain Adaptation.

Data-Adaptive Metric Learning with Scale Alignment.

Domain Adaptation Based on Semi-Supervised Cross-Domain Mean Discriminative Analysis and Kernel Transfer Extreme Learning Machine

Towards Self-Similarity Consistency and Feature Discrimination for Unsupervised Domain Adaptation.

Deep Discriminative Domain Adaptation.

A Visual Domain Adaptation Method Based on Enhanced Subspace Distribution Matching

Enhanced Subspace Distribution Matching for Fast Visual Domain Adaptation

A Unified Joint Maximum Mean Discrepancy for Domain Adaptation

A Risk Minimization Framework for Domain Adaptation

Joint Metric and Feature Representation Learning for Unsupervised Domain Adaptation

Coupled locality discriminant analysis with globality preserving for dimensionality reduction

Joint Adaptive Dual Graph and Feature Selection for Domain Adaptation

Discriminative and Geometry-Aware Unsupervised Domain Adaptation

Deep Domain Similarity Adaptation Networks for Across Domain Classification

Domain Invariant and Class Discriminative Feature Learning for Visual Domain Adaptation