Learning Transferable Features with Deep Adaptation Networks

2015/02/10 by Mingsheng Long, Yue Cao, Long, Mingsheng +5 · 97 citations
Computer Science · #Anomaly Detection Techniques and Applications #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis #Machine Learning (cs.LG)

paper · pdf · doi:10.48550/arxiv.1502.02791

openalex publication_date 2015/02/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Recent studies reveal that a deep neural network can learn transferable features which generalize well to novel tasks for domain adaptation. However, as deep features eventually transition from general to specific along the network, the feature transferability drops significantly in higher layers with increasing domain discrepancy. Hence, it is important to formally reduce the dataset bias and enhance the transferability in task-specific layers. In this paper, we propose a new Deep Adaptation Network (DAN) architecture, which generalizes deep convolutional neural network to the domain adaptation scenario. In DAN, hidden representations of all task-specific layers are embedded in a reproducing kernel Hilbert space where the mean embeddings of different domain distributions can be explicitly matched. The domain discrepancy is further reduced using an optimal multi-kernel selection method for mean embedding matching. DAN can learn transferable features with statistical guarantees, and can scale linearly by unbiased estimate of kernel embedding. Extensive empirical evidence shows that the proposed architecture yields state-of-the-art image classification error rates on standard domain adaptation benchmarks.

Citations

• Deep Domain Confusion: Maximizing for Domain Invariance
• Domain-Adversarial Neural Networks
• Domain Adaptation: Learning Bounds and Algorithms
• Improving neural networks by preventing co-adaptation of feature detectors
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual\n Recognition
• Representation Learning: A Review and New Perspectives

Cited by

• Exploring Syn-to-Real Domain Adaptation for Military Target Detection
• Source-Free Controlled Adaptation of Teachers for Continual Test-Time Adaptation
• Breaking the Synthetic-Real Domain Shortcut for Training-Free Generative Replay-based Class Incremental Learning
• CD-RMOT-Bench: Benchmarking the Cross-Domain Referring Multi-Object Tracking
• C-DGPA: Class-Centric Dual-Alignment Generative Prompt Adaptation
• Enhancing Node-Level Graph Domain Adaptation by Alleviating Local Dependency
• Federated Domain Generalization with Latent Space Inversion
• Balanced Learning for Domain Adaptive Semantic Segmentation
• Variance Matters: Improving Domain Adaptation via Stratified Sampling
• Label-Efficient Hyperspectral Image Classification via Spectral FiLM Modulation of Low-Level Pretrained Diffusion Features
• Bridging Modalities via Progressive Re-alignment for Multimodal Test-Time Adaptation
• LFreeDA: Label-Free Drift Adaptation for Windows Malware Detection
• Data Value in the Age of Scaling: Understanding LLM Scaling Dynamics Under Real-Synthetic Data Mixtures
• Unsupervised Robust Domain Adaptation: Paradigm, Theory and Algorithm
• Boomda: Balanced Multi-objective Optimization for Multimodal Domain Adaptation
• TCSA-UDA: Text-Driven Cross-Semantic Alignment for Unsupervised Domain Adaptation in Medical Image Segmentation
• Machine and Deep Learning for Indoor UWB Jammer Localization
• PF-DAformer: Proximal Femur Segmentation via Domain Adaptive Transformer for Dual-Center QCT
• Using latent representations to link disjoint longitudinal data for mixed-effects regression
• Instance-Aware Pseudo-Labeling and Class-Focused Contrastive Learning for Weakly Supervised Domain Adaptive Segmentation of Electron Microscopy
• Mitigating Negative Transfer via Reducing Environmental Disagreement
• PSScreen V2: Partially Supervised Multiple Retinal Disease Screening
• GALA: A GlobAl-LocAl Approach for Multi-Source Active Domain Adaptation
• Attention Residual Fusion Network with Contrast for Source-free Domain Adaptation
• Adaptive Split-MMD Training for Small-Sample Cross-Dataset P300 EEG Classification
• H-SPLID: HSIC-based Saliency Preserving Latent Information Decomposition
• Deep Learning Based Domain Adaptation Methods in Remote Sensing: A Comprehensive Survey
• Geometric Moment Alignment for Domain Adaptation via Siegel Embeddings
• High-Fidelity Simulated Data Generation for Real-World Zero-Shot Robotic Manipulation Learning with Gaussian Splatting
• Reverse Supervision at Scale: Exponential Search Meets the Economics of Annotation
• FRIEREN: Federated Learning with Vision-Language Regularization for Segmentation
• VirDA: Reusing Backbone for Unsupervised Domain Adaptation with Visual Reprogramming
• Approximate Domain Unlearning for Vision-Language Models
• Consistent Assistant Domains Transformer for Source-free Domain Adaptation
• Deep Domain Adaptation for Turbofan Engine Remaining Useful Life Prediction: Methodologies, Evaluation and Future Trends
• Multi-Level Heterogeneous Knowledge Transfer Network on Forward Scattering Center Model for Limited Samples SAR ATR
• POEM: Explore Unexplored Reliable Samples to Enhance Test-Time Adaptation
• Unsupervised Domain Adaptation for Binary Classification with an Unobservable Source Subpopulation
• Towards Practical Algorithm Selection for Unsupervised Domain Adaptation in Medical Imaging
• Online Adaptation via Dual-Stage Alignment and Self-Supervision for Fast-Calibration Brain-Computer Interfaces
• Generalizable Domain Adaptation for Sim-and-Real Policy Co-Training
• COLA: Context-aware Language-driven Test-time Adaptation
• Training-Free Label Space Alignment for Universal Domain Adaptation
• Lost in Translation? Vocabulary Alignment for Source-Free Adaptation in Open-Vocabulary Semantic Segmentation
• Temporal Attentive Alignment for Large-Scale Video Domain Adaptation
• Unsupervised Domain Adaptation of Black-Box Source Models
• Domain Adaptation with Auxiliary Target Domain-Oriented Classifier
• Sparse-to-dense Feature Matching: Intra and Inter domain Cross-modal Learning in Domain Adaptation for 3D Semantic Segmentation
• Learning Majority-to-Minority Transformations with MMD and Triplet Loss for Imbalanced Classification
• "A 6 or a 9?": Ensemble Learning Through the Multiplicity of Performant Models and Explanations
• Learning Invariant Representation with Consistency and Diversity for Semi-supervised Source Hypothesis Transfer
• CDTrans: Cross-domain Transformer for Unsupervised Domain Adaptation
• Self-Supervised Training Enhances Online Continual Learning
• ACE and Diverse Generalization via Selective Disagreement
• What Are We Automating? On the Need for Vision and Expertise When Deploying AI Systems
• Semi-supervised Deep Transfer for Regression without Domain Alignment
• Leveraging Transfer Learning and Mobile-enabled Convolutional Neural Networks for Improved Arabic Handwritten Character Recognition
• LogME: Practical Assessment of Pre-trained Models for Transfer Learning
• WILDS: A Benchmark of in-the-Wild Distribution Shifts
• Online Meta-Learning for Multi-Source and Semi-Supervised Domain Adaptation
• Empowering Large Language Model for Sequential Recommendation via Multimodal Embeddings and Semantic IDs
• LoveDA: A Remote Sensing Land-Cover Dataset for Domain Adaptive Semantic Segmentation
• Domain Attention Consistency for Multi-Source Domain Adaptation
• A Review of Single-Source Deep Unsupervised Visual Domain Adaptation
• Normalized Wasserstein Distance for Mixture Distributions with Applications in Adversarial Learning and Domain Adaptation
• Unsupervised Domain Adaptation via Discriminative Manifold Embedding and Alignment
• Contrastively Smoothed Class Alignment for Unsupervised Domain Adaptation
• DINE: Domain Adaptation from Single and Multiple Black-box Predictors
• Teacher Model Fingerprinting Attacks Against Transfer Learning
• Exploring Categorical Regularization for Domain Adaptive Object Detection
• Cross-Domain Gradient Discrepancy Minimization for Unsupervised Domain Adaptation
• FixBi: Bridging Domain Spaces for Unsupervised Domain Adaptation
• Discriminative Noise Robust Sparse Orthogonal Label Regression-based Domain Adaptation
• Entropy Minimization vs. Diversity Maximization for Domain Adaptation
• Open Set Domain Adaptation for Image and Action Recognition
• Ranking and Tuning Pre-trained Models: A New Paradigm for Exploiting Model Hubs
• Cross-Cancer Knowledge Transfer in WSI-based Prognosis Prediction
• Cross-domain Object Detection through Coarse-to-Fine Feature Adaptation
• Unsupervised Domain Adaptation for Semantic Segmentation via Low-level Edge Information Transfer
• Wasserstein Distance Guided Cross-Domain Learning
• Generalizable Federated Learning using Client Adaptive Focal Modulation
• PSScreen: Partially Supervised Multiple Retinal Disease Screening
• How Safe Will I Be Given What I Saw? Calibrated Prediction of Safety Chances for Image-Controlled Autonomy
• Deep Symmetric Adaptation Network for Cross-modality Medical Image Segmentation
• Prototypical Cross-domain Self-supervised Learning for Few-shot Unsupervised Domain Adaptation
• Unsupervised Domain Alignment to Mitigate Low Level Dataset Biases
• More Separable and Easier to Segment: A Cluster Alignment Method for Cross-Domain Semantic Segmentation
• Unified modality separation: A vision-language framework for unsupervised domain adaptation
• SPA++: Generalized Graph Spectral Alignment for Versatile Domain Adaptation
• Slice or the Whole Pie? Utility Control for AI Models
• CORE-ReID V2: Advancing the Domain Adaptation for Object Re-Identification with Optimized Training and Ensemble Fusion
• Neutralizing Token Aggregation via Information Augmentation for Efficient Test-Time Adaptation
• DART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
• Domain Adaptation with Incomplete Target Domains
• Nested Graph Pseudo-Label Refinement for Noisy Label Domain Adaptation Learning
• Consensus-Driven Active Model Selection
• Secure Domain Adaptation with Multiple Sources

Related

• Deep Transfer Network: Unsupervised Domain Adaptation
• Unsupervised Domain Adaptation with Residual Transfer Networks
• Network Architecture Search for Domain Adaptation
• Simultaneous Deep Transfer Across Domains and Tasks
• Incremental Learning Through Deep Adaptation