Aligning Domain-specific Distribution and Classifier for Cross-domain Classification from Multiple Sources

Abstract

While Unsupervised Domain Adaptation (UDA) algorithms, i.e., there are only labeled data from source domains, have been actively studied in recent years, most algorithms and theoretical results focus on Single-source Unsupervised Domain Adaptation (SUDA). However, in the practical scenario, labeled data can be typically collected from multiple diverse sources, and they might be different not only from the target domain but also from each other. Thus, domain adapters from multiple sources should not be modeled in the same way. Recent deep learning based Multi-source Unsupervised Domain Adaptation (MUDA) algorithms focus on extracting common domain-invariant representations for all domains by aligning distribution of all pairs of source and target domains in a common feature space. However, it is often very hard to extract the same domain-invariant representations for all domains in MUDA. In addition, these methods match distributions without considering domain-specific decision boundaries between classes. To solve these problems, we propose a new framework with two alignment stages for MUDA which not only respectively aligns the distributions of each pair of source and target domains in multiple specific feature spaces, but also aligns the outputs of classifiers by utilizing the domain-specific decision boundaries. Extensive experiments demonstrate that our method can achieve remarkable results on popular benchmark datasets for image classification.

arXiv commentAAAI 2019 long paper. Multi-source Domain Adaptation

References in corpus (10)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
• Learning Transferable Features with Deep Adaptation Networks
• Unsupervised Domain Adaptation by Backpropagation
• Deep Domain Confusion: Maximizing for Domain Invariance
• Unsupervised Domain Adaptation with Residual Transfer Networks
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• Deep CORAL: Correlation Alignment for Deep Domain Adaptation
• Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
• Deep Cocktail Network: Multi-source Unsupervised Domain Adaptation with Category Shift

Cited by in corpus (27)

• Deep Subdomain Adaptation Network for Image Classification
• Multi-Representation Adaptation Network for Cross-domain Image Classification
• MDFEND: Multi-domain Fake News Detection
• A Comprehensive Survey on Transfer Learning
• Memory-Guided Multi-View Multi-Domain Fake News Detection
• Multi-Source Unsupervised Domain Adaptation via Pseudo Target Domain
• FastSurferVINN: Building Resolution-Independence into Deep Learning Segmentation Methods -- A Solution for HighRes Brain MRI
• Multi-source Domain Adaptation in the Deep Learning Era: A Systematic Survey
• Modeling Users' Behavior Sequences with Hierarchical Explainable Network for Cross-domain Fraud Detection
• Robust Domain Misinformation Detection via Multi-modal Feature Alignment
• Multi-source Domain Adaptation for Text-independent Forensic Speaker Recognition
• A Prototype-Oriented Framework for Unsupervised Domain Adaptation
• Multi-Source EEG Emotion Recognition via Dynamic Contrastive Domain Adaptation
• Domain Attention Consistency for Multi-Source Domain Adaptation
• DINE: Domain Adaptation from Single and Multiple Black-box Predictors
• Aggregating From Multiple Target-Shifted Sources
• SCIDA: Self-Correction Integrated Domain Adaptation from Single- to Multi-label Aerial Images
• Secure Domain Adaptation with Multiple Sources
• Stacking Ensemble Learning in Deep Domain Adaptation for Ophthalmic Image Classification
• AED-PADA:Improving Generalizability of Adversarial Example Detection via Principal Adversarial Domain Adaptation
• Dynamic Feature Alignment for Semi-supervised Domain Adaptation
• Multi-source Heterogeneous Domain Adaptation with Conditional Weighting Adversarial Network
• Multi-source Few-shot Domain Adaptation
• DRR4Covid: Learning Automated COVID-19 Infection Segmentation from Digitally Reconstructed Radiographs
• Universal Multi-Source Domain Adaptation
• Domain Consistency Regularization for Unsupervised Multi-source Domain Adaptive Classification
• Cross-domain Adaptation with Discrepancy Minimization for Text-independent Forensic Speaker Verification