Domain Separation Networks

Abstract

The cost of large scale data collection and annotation often makes the application of machine learning algorithms to new tasks or datasets prohibitively expensive. One approach circumventing this cost is training models on synthetic data where annotations are provided automatically. Despite their appeal, such models often fail to generalize from synthetic to real images, necessitating domain adaptation algorithms to manipulate these models before they can be successfully applied. Existing approaches focus either on mapping representations from one domain to the other, or on learning to extract features that are invariant to the domain from which they were extracted. However, by focusing only on creating a mapping or shared representation between the two domains, they ignore the individual characteristics of each domain. We suggest that explicitly modeling what is unique to each domain can improve a model's ability to extract domain-invariant features. Inspired by work on private-shared component analysis, we explicitly learn to extract image representations that are partitioned into two subspaces: one component which is private to each domain and one which is shared across domains. Our model is trained not only to perform the task we care about in the source domain, but also to use the partitioned representation to reconstruct the images from both domains. Our novel architecture results in a model that outperforms the state-of-the-art on a range of unsupervised domain adaptation scenarios and additionally produces visualizations of the private and shared representations enabling interpretation of the domain adaptation process.

This work will be presented at NIPS 2016

References in corpus (5)

• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• Learning Transferable Features with Deep Adaptation Networks
• Deep Domain Confusion: Maximizing for Domain Invariance
• Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
• Domain-Adversarial Neural Networks

Cited by in corpus (244)

• Deep Visual Domain Adaptation: A Survey
• A review of domain adaptation without target labels
• Model Adaptation: Unsupervised Domain Adaptation without Source Data
• Beyond Sharing Weights for Deep Domain Adaptation
• Asymmetric Tri-training for Unsupervised Domain Adaptation
• A DIRT-T Approach to Unsupervised Domain Adaptation
• Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification
• Revisiting Batch Normalization For Practical Domain Adaptation
• Multimodal Unsupervised Image-to-Image Translation
• Cross-view Semantic Segmentation for Sensing Surroundings
• Learning Robust Global Representations by Penalizing Local Predictive Power
• Weakly Supervised Adversarial Domain Adaptation for Semantic Segmentation in Urban Scenes
• Task Driven Generative Modeling for Unsupervised Domain Adaptation: Application to X-ray Image Segmentation
• DRIT++: Diverse Image-to-Image Translation via Disentangled Representations
• Unsupervised Domain Adaptation through Self-Supervision
• Deep Residual Correction Network for Partial Domain Adaptation
• On Learning Invariant Representation for Domain Adaptation
• Automatic Sleep Staging of EEG Signals: Recent Development, Challenges, and Future Directions
• Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
• Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation
• Speaker-Invariant Training via Adversarial Learning
• Deep Transfer Learning with Joint Adaptation Networks
• Contrastive Adaptation Network for Unsupervised Domain Adaptation
• Robust Unsupervised Domain Adaptation for Neural Networks via Moment Alignment
• Adversarial Multi-task Learning for Text Classification
• Classification of crystallization outcomes using deep convolutional neural networks
• Unsupervised Sound Separation Using Mixture Invariant Training
• Multi-Adversarial Domain Adaptation
• Adversarial Teacher-Student Learning for Unsupervised Domain Adaptation
• Image-to-image translation for cross-domain disentanglement
• Semi-supervised Domain Adaptation via Minimax Entropy
• Cross-Domain Complementary Learning Using Pose for Multi-Person Part Segmentation
• Feature-Critic Networks for Heterogeneous Domain Generalization
• Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks
• Unsupervised Adaptation with Domain Separation Networks for Robust Speech Recognition
• Pseudo-Labeling Curriculum for Unsupervised Domain Adaptation
• Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift
• Maximum-Entropy Adversarial Data Augmentation for Improved Generalization and Robustness
• Multi-modal Deep Analysis for Multimedia
• Domain Adaptation with Asymmetrically-Relaxed Distribution Alignment
• COVID-DA: Deep Domain Adaptation from Typical Pneumonia to COVID-19
• MISA: Modality-Invariant and -Specific Representations for Multimodal Sentiment Analysis
• Invariance Matters: Exemplar Memory for Domain Adaptive Person Re-identification
• Domain Adaptation for Underwater Image Enhancement via Content and Style Separation
• Domain Adaptation for Semantic Segmentation via Class-Balanced Self-Training
• Moment Matching for Multi-Source Domain Adaptation
• Virtual Mixup Training for Unsupervised Domain Adaptation
• Larger Norm More Transferable: An Adaptive Feature Norm Approach for Unsupervised Domain Adaptation
• Contrastive Variational Autoencoder Enhances Salient Features
• Multi-reference Tacotron by Intercross Training for Style Disentangling,Transfer and Control in Speech Synthesis
• Support and Invertibility in Domain-Invariant Representations
• Adversarial Speaker Adaptation
• Exploring Domain Shift in Extractive Text Summarization
• Inferring Latent Domains for Unsupervised Deep Domain Adaptation
• Learning to Adapt Invariance in Memory for Person Re-identification
• Meta-FDMixup: Cross-Domain Few-Shot Learning Guided by Labeled Target Data
• Adversarial Multi-Criteria Learning for Chinese Word Segmentation
• Domain Adaptive Attention Learning for Unsupervised Person Re-Identification
• Drop to Adapt: Learning Discriminative Features for Unsupervised Domain Adaptation
• An Annotation Saved is an Annotation Earned: Using Fully Synthetic Training for Object Instance Detection
• Source Free Domain Adaptation with Image Translation
• 3D Object Detection and Pose Estimation of Unseen Objects in Color Images with Local Surface Embeddings
• Source-free Domain Adaptation via Distributional Alignment by Matching Batch Normalization Statistics
• Multiple Source Domain Adaptation with Adversarial Training of Neural Networks
• Joint Domain Alignment and Discriminative Feature Learning for Unsupervised Deep Domain Adaptation
• Transfer of Machine Learning Fairness across Domains
• Episodic Training for Domain Generalization
• Putting Fairness Principles into Practice: Challenges, Metrics, and Improvements
• SPIGAN: Privileged Adversarial Learning from Simulation
• Domain Aggregation Networks for Multi-Source Domain Adaptation
• Orthogonal Language and Task Adapters in Zero-Shot Cross-Lingual Transfer
• Domain-Specific Batch Normalization for Unsupervised Domain Adaptation
• Source Data-absent Unsupervised Domain Adaptation through Hypothesis Transfer and Labeling Transfer
• Domain Adaptation without Source Data
• Progressive Feature Alignment for Unsupervised Domain Adaptation
• A Review of Single-Source Deep Unsupervised Visual Domain Adaptation
• Review-Based Domain Disentanglement without Duplicate Users or Contexts for Cross-Domain Recommendation
• DCAN: Dual Channel-wise Alignment Networks for Unsupervised Scene Adaptation
• Open Set Domain Adaptation by Backpropagation
• Causal Generative Domain Adaptation Networks
• Geometry-Consistent Generative Adversarial Networks for One-Sided Unsupervised Domain Mapping
• A Novel Unsupervised Camera-aware Domain Adaptation Framework for Person Re-identification
• Domain Adaptation for Time Series Forecasting via Attention Sharing
• DARec: Deep Domain Adaptation for Cross-Domain Recommendation via Transferring Rating Patterns
• A Broader Study of Cross-Domain Few-Shot Learning
• Image Generation From Small Datasets via Batch Statistics Adaptation
• Deep Adversarial Attention Alignment for Unsupervised Domain Adaptation: the Benefit of Target Expectation Maximization
• Domain-Invariant Adversarial Learning for Unsupervised Domain Adaption
• Gradually Vanishing Bridge for Adversarial Domain Adaptation
• CANU-ReID: A Conditional Adversarial Network for Unsupervised person Re-IDentification
• Entropy Minimization vs. Diversity Maximization for Domain Adaptation
• From source to target and back: symmetric bi-directional adaptive GAN
• LiDARNet: A Boundary-Aware Domain Adaptation Model for Point Cloud Semantic Segmentation
• Progressive Domain Expansion Network for Single Domain Generalization
• Modified Distribution Alignment for Domain Adaptation with Pre-trained Inception ResNet
• Learning Factorized Representations for Open-set Domain Adaptation
• CLDA: Contrastive Learning for Semi-Supervised Domain Adaptation
• Aspect-augmented Adversarial Networks for Domain Adaptation
• Learning to Learn with Variational Information Bottleneck for Domain Generalization
• Cross-domain Recommendation via Deep Domain Adaptation
• Dynamic Weighted Learning for Unsupervised Domain Adaptation
• Anomaly Detection with Domain Adaptation
• Domain Adaptive Adversarial Learning Based on Physics Model Feedback for Underwater Image Enhancement
• Network Architecture Search for Domain Adaptation
• Learning Invariant Representations and Risks for Semi-supervised Domain Adaptation
• Adversarial Alignment of Class Prediction Uncertainties for Domain Adaptation
• Attentive Adversarial Learning for Domain-Invariant Training
• CAN: Constrained Attention Networks for Multi-Aspect Sentiment Analysis
• Multi-Task Domain Adaptation for Deep Learning of Instance Grasping from Simulation
• One Size Does Not Fit All: Generating and Evaluating Variable Number of Keyphrases
• Towards Stable and Comprehensive Domain Alignment: Max-Margin Domain-Adversarial Training
• DART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
• Exploiting Coarse-to-Fine Task Transfer for Aspect-level Sentiment Classification
• Self-Supervised Sim-to-Real Adaptation for Visual Robotic Manipulation
• Two-stage Temporal Modelling Framework for Video-based Depression Recognition using Graph Representation
• Domain Adaptation with Auxiliary Target Domain-Oriented Classifier
• Contrastively Smoothed Class Alignment for Unsupervised Domain Adaptation
• Aggregate or Not? Exploring Where to Privatize in DNN Based Federated Learning Under Different Non-IID Scenes
• Adversarial Camera Alignment Network for Unsupervised Cross-camera Person Re-identification
• SCIDA: Self-Correction Integrated Domain Adaptation from Single- to Multi-label Aerial Images
• Adaptation Across Extreme Variations using Unlabeled Domain Bridges
• Cross-Domain Perceptual Reward Functions
• Multi-source Distilling Domain Adaptation
• Conditional Adversarial Networks for Multi-Domain Text Classification
• Class-imbalanced Domain Adaptation: An Empirical Odyssey
• Domain-adaptive Crowd Counting via High-quality Image Translation and Density Reconstruction
• Deep Adversarial Transition Learning using Cross-Grafted Generative Stacks
• Adaptation and Re-Identification Network: An Unsupervised Deep Transfer Learning Approach to Person Re-Identification
• Multi-Source Cross-Lingual Model Transfer: Learning What to Share
• Unsupervised Visual Domain Adaptation: A Deep Max-Margin Gaussian Process Approach
• Attention-Privileged Reinforcement Learning
• Learning Smooth Representation for Unsupervised Domain Adaptation
• Exclusive Hierarchical Decoding for Deep Keyphrase Generation
• Online Meta-Learning for Multi-Source and Semi-Supervised Domain Adaptation
• CMTS: Conditional Multiple Trajectory Synthesizer for Generating Safety-critical Driving Scenarios
• Domain-Adaptive Few-Shot Learning
• GA-DAN: Geometry-Aware Domain Adaptation Network for Scene Text Detection and Recognition
• Discriminative Adversarial Domain Adaptation
• ADAIL: Adaptive Adversarial Imitation Learning
• Powers of layers for image-to-image translation
• Unsupervised Domain Adaptation Schemes for Building ASR in Low-resource Languages
• A Cross-Level Information Transmission Network for Predicting Phenotype from New Genotype: Application to Cancer Precision Medicine
• Towards Backward-Compatible Representation Learning
• Pixel and Feature Level Based Domain Adaption for Object Detection in Autonomous Driving
• Factorized Adversarial Networks for Unsupervised Domain Adaptation
• SHAPED: Shared-Private Encoder-Decoder for Text Style Adaptation
• Emotion controllable speech synthesis using emotion-unlabeled dataset with the assistance of cross-domain speech emotion recognition
• Target-Embedding Autoencoders for Supervised Representation Learning
• Learning to Diversify for Single Domain Generalization
• EFANet: Exchangeable Feature Alignment Network for Arbitrary Style Transfer
• Towards Disentangled Representations for Human Retargeting by Multi-view Learning
• Moment-Based Domain Adaptation: Learning Bounds and Algorithms
• Joint Semantic Domain Alignment and Target Classifier Learning for Unsupervised Domain Adaptation
• User-specific Adaptive Fine-tuning for Cross-domain Recommendations
• Graceful Degradation and Related Fields
• Multi-Source Domain Adaptation with Mixture of Experts
• MADAN: Multi-source Adversarial Domain Aggregation Network for Domain Adaptation
• Towards Discriminative Representation Learning for Unsupervised Person Re-identification
• DRANet: Disentangling Representation and Adaptation Networks for Unsupervised Cross-Domain Adaptation
• Driving Experience Transfer Method for End-to-End Control of Self-Driving Cars
• Unsupervised Cross-domain Image Classification by Distance Metric Guided Feature Alignment
• Deep Symmetric Adaptation Network for Cross-modality Medical Image Segmentation
• Generalizable Model-agnostic Semantic Segmentation via Target-specific Normalization
• Domain Adaptation for Sentiment Analysis Using Increased Intraclass Separation
• CUDA: Contradistinguisher for Unsupervised Domain Adaptation
• Rare Event Detection using Disentangled Representation Learning
• CNN-based Discriminative Training for Domain Compensation in Acoustic Event Detection with Frame-wise Classifier
• Unsupervised Domain Adaptation using Generative Models and Self-ensembling
• Fairness in TabNet Model by Disentangled Representation for the Prediction of Hospital No-Show
• Locality Preserving Joint Transfer for Domain Adaptation
• AdaFlow: Domain-Adaptive Density Estimator with Application to Anomaly Detection and Unpaired Cross-Domain Translation
• CODE-AE: A Coherent De-confounding Autoencoder for Predicting Patient-Specific Drug Response From Cell Line Transcriptomics
• Multi-Source Domain Adaptation for Text Classification via DistanceNet-Bandits
• Heterogeneous Domain Adaptation via Soft Transfer Network
• Domain Adaptation Meets Disentangled Representation Learning and Style Transfer
• Visualizing Adapted Knowledge in Domain Transfer
• Dual Adversarial Co-Learning for Multi-Domain Text Classification
• Gotta Adapt 'Em All: Joint Pixel and Feature-Level Domain Adaptation for Recognition in the Wild
• Generative Adversarial Networks are Special Cases of Artificial Curiosity (1990) and also Closely Related to Predictability Minimization (1991)
• Domain-Adversarial and Conditional State Space Model for Imitation Learning
• Multiple Classifiers Based Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
• Self-adaptive Re-weighted Adversarial Domain Adaptation
• Cross-Domain Adversarial Auto-Encoder
• Learning from Synthetic Animals
• Spatial-Aware GAN for Unsupervised Person Re-identification
• Discovering Differential Features: Adversarial Learning for Information Credibility Evaluation
• Domain-Invariant Feature Distillation for Cross-Domain Sentiment Classification
• Artistic Domain Generalisation Methods are Limited by their Deep Representations
• Preliminary Steps Towards Federated Sentiment Classification
• VLAD-VSA: Cross-Domain Face Presentation Attack Detection with Vocabulary Separation and Adaptation
• Transporting Causal Mechanisms for Unsupervised Domain Adaptation
• Sequential Learning for Domain Generalization
• Adversarial Training Based Multi-Source Unsupervised Domain Adaptation for Sentiment Analysis
• Unsupervised Multi-Domain Multimodal Image-to-Image Translation with Explicit Domain-Constrained Disentanglement
• Impact of ImageNet Model Selection on Domain Adaptation
• Adversarial Adaptation of Scene Graph Models for Understanding Civic Issues
• Unsupervised Domain Adaptation using Deep Networks with Cross-Grafted Stacks
• Structured Discriminative Tensor Dictionary Learning for Unsupervised Domain Adaptation
• Incremental Unsupervised Domain-Adversarial Training of Neural Networks
• Enlarging Discriminative Power by Adding an Extra Class in Unsupervised Domain Adaptation
• Multi-Facet Recommender Networks with Spherical Optimization
• Divergence Optimization for Noisy Universal Domain Adaptation
• Efficient Multi-Domain Network Learning by Covariance Normalization
• Explainability-aided Domain Generalization for Image Classification
• Synthetic Data Generation for Classifying Electrophysiological and Morpho-Electrophysiological Neurons from Mouse Visual Cortex
• Multi-Purposing Domain Adaptation Discriminators for Pseudo Labeling Confidence
• Semi-Supervised Adversarial Discriminative Domain Adaptation
• Weighed Domain-Invariant Representation Learning for Cross-domain Sentiment Analysis
• Unsupervised Sentiment Analysis by Transferring Multi-source Knowledge
• Learning to adapt class-specific features across domains for semantic segmentation
• Empirical Evaluation of Multi-task Learning in Deep Neural Networks for Natural Language Processing
• Semantically Robust Unpaired Image Translation for Data with Unmatched Semantics Statistics
• Deep Shape Matching
• Dynamic Layer Customization for Noise Robust Speech Emotion Recognition in Heterogeneous Condition Training
• A Dictionary Approach to Domain-Invariant Learning in Deep Networks
• Deep Discriminative Learning for Unsupervised Domain Adaptation
• Nonlinear Collaborative Scheme for Deep Neural Networks
• Diverse Adversaries for Mitigating Bias in Training
• Discriminative Noise Robust Sparse Orthogonal Label Regression-based Domain Adaptation
• Towards Recognizing New Semantic Concepts in New Visual Domains
• Mixup Regularized Adversarial Networks for Multi-Domain Text Classification
• The Role of Embedding Complexity in Domain-invariant Representations
• Wasserstein Distance Guided Cross-Domain Learning
• Domain-Adversarial Multi-Task Framework for Novel Therapeutic Property Prediction of Compounds
• Localising In Complex Scenes Using Balanced Adversarial Adaptation
• Weakly Supervised Domain Detection
• Discriminative Domain-Invariant Adversarial Network for Deep Domain Generalization
• Domain-robust VQA with diverse datasets and methods but no target labels
• Leveraging Motion Priors in Videos for Improving Human Segmentation
• Phonetic-and-Semantic Embedding of Spoken Words with Applications in Spoken Content Retrieval
• Shared Predictive Cross-Modal Deep Quantization
• Deep Domain Generalization with Feature-norm Network
• Domain Agnostic Learning for Unbiased Authentication
• Learning to see across Domains and Modalities
• Adversarial Domain Randomization
• Contrastive Vicinal Space for Unsupervised Domain Adaptation
• CHEER: Rich Model Helps Poor Model via Knowledge Infusion
• Domain Adaptation with Incomplete Target Domains
• Modality Compensation Network: Cross-Modal Adaptation for Action Recognition
• Synthetic-to-Real Domain Adaptation for Lane Detection
• Decentralized Deep Reinforcement Learning for Network Level Traffic Signal Control
• Learning Task-oriented Disentangled Representations for Unsupervised Domain Adaptation
• Cross-Domain Latent Modulation for Variational Transfer Learning
• Learning to Generalize One Sample at a Time with Self-Supervision