Adversarial Discriminative Domain Adaptation

Abstract

Adversarial learning methods are a promising approach to training robust deep networks, and can generate complex samples across diverse domains. They also can improve recognition despite the presence of domain shift or dataset bias: several adversarial approaches to unsupervised domain adaptation have recently been introduced, which reduce the difference between the training and test domain distributions and thus improve generalization performance. Prior generative approaches show compelling visualizations, but are not optimal on discriminative tasks and can be limited to smaller shifts. Prior discriminative approaches could handle larger domain shifts, but imposed tied weights on the model and did not exploit a GAN-based loss. We first outline a novel generalized framework for adversarial adaptation, which subsumes recent state-of-the-art approaches as special cases, and we use this generalized view to better relate the prior approaches. We propose a previously unexplored instance of our general framework which combines discriminative modeling, untied weight sharing, and a GAN loss, which we call Adversarial Discriminative Domain Adaptation (ADDA). We show that ADDA is more effective yet considerably simpler than competing domain-adversarial methods, and demonstrate the promise of our approach by exceeding state-of-the-art unsupervised adaptation results on standard cross-domain digit classification tasks and a new more difficult cross-modality object classification task.

References in corpus (8)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Conditional Generative Adversarial Nets
• Caffe: Convolutional Architecture for Fast Feature Embedding
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
• Unsupervised Domain Adaptation by Backpropagation
• Deep Domain Confusion: Maximizing for Domain Invariance
• Beyond Sharing Weights for Deep Domain Adaptation
• Deep CORAL: Correlation Alignment for Deep Domain Adaptation

Cited by in corpus (180)

• Deep Visual Domain Adaptation: A Survey
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification
• Generalizing to Unseen Domains via Adversarial Data Augmentation
• Like What You Like: Knowledge Distill via Neuron Selectivity Transfer
• Generalizing Across Domains via Cross-Gradient Training
• Domain Adaptation for Visual Applications: A Comprehensive Survey
• Multi-Representation Adaptation Network for Cross-domain Image Classification
• Tent: Fully Test-time Adaptation by Entropy Minimization
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Few-Shot Adversarial Domain Adaptation
• A Survey of Domain Adaptation for Neural Machine Translation
• The Numerics of GANs
• On Learning Invariant Representation for Domain Adaptation
• Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
• Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation
• Learning Disentangled Semantic Representation for Domain Adaptation
• Learning to cluster in order to transfer across domains and tasks
• Deep Transfer Learning with Joint Adaptation Networks
• Discovering New Strong Gravitational Lenses in the DESI Legacy Imaging Surveys
• Semi-supervised Domain Adaptation via Minimax Entropy
• Semantic-aware Grad-GAN for Virtual-to-Real Urban Scene Adaption
• Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift
• Adversarial Style Mining for One-Shot Unsupervised Domain Adaptation
• Sample-to-Sample Correspondence for Unsupervised Domain Adaptation
• Fully Convolutional Adaptation Networks for Semantic Segmentation
• Self-ensembling for visual domain adaptation
• Controllable Invariance through Adversarial Feature Learning
• Label Efficient Learning of Transferable Representations across Domains and Tasks
• Deep Cocktail Network: Multi-source Unsupervised Domain Adaptation with Category Shift
• Adversarial Dropout Regularization
• Transferrable Prototypical Networks for Unsupervised Domain Adaptation
• Vertex and Energy Reconstruction in JUNO with Machine Learning Methods
• Mid-Level Visual Representations Improve Generalization and Sample Efficiency for Learning Visuomotor Policies
• PnP-AdaNet: Plug-and-Play Adversarial Domain Adaptation Network with a Benchmark at Cross-modality Cardiac Segmentation
• Minimal-Entropy Correlation Alignment for Unsupervised Deep Domain Adaptation
• Domain Adaptation for Semantic Segmentation with Maximum Squares Loss
• A Survey on Deep Transfer Learning
• CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise
• Invariant Causal Prediction for Block MDPs
• Multi-target Unsupervised Domain Adaptation without Exactly Shared Categories
• Unsupervised Cross-Modality Domain Adaptation of ConvNets for Biomedical Image Segmentations with Adversarial Loss
• Exploring Object Relation in Mean Teacher for Cross-Domain Detection
• Drop to Adapt: Learning Discriminative Features for Unsupervised Domain Adaptation
• Partial Adversarial Domain Adaptation
• Unseen Target Stance Detection with Adversarial Domain Generalization
• Joint Domain Alignment and Discriminative Feature Learning for Unsupervised Deep Domain Adaptation
• Decentralized Learning of Generative Adversarial Networks from Non-iid Data
• Image to Image Translation for Domain Adaptation
• Knowledge Distillation Methods for Efficient Unsupervised Adaptation Across Multiple Domains
• Significance-aware Information Bottleneck for Domain Adaptive Semantic Segmentation
• Harmonizing Transferability and Discriminability for Adapting Object Detectors
• DCAN: Dual Channel-wise Alignment Networks for Unsupervised Scene Adaptation
• Open Set Domain Adaptation by Backpropagation
• On Learning Associations of Faces and Voices
• Importance Weighted Adversarial Nets for Partial Domain Adaptation
• Zero-Shot Visual Recognition using Semantics-Preserving Adversarial Embedding Networks
• Joint Disentangling and Adaptation for Cross-Domain Person Re-Identification
• All about Structure: Adapting Structural Information across Domains for Boosting Semantic Segmentation
• Overcoming Language Priors in Visual Question Answering with Adversarial Regularization
• Frustratingly Easy Person Re-Identification: Generalizing Person Re-ID in Practice
• Jointly De-biasing Face Recognition and Demographic Attribute Estimation
• Domain Adversarial Training for Accented Speech Recognition
• Image Generation From Small Datasets via Batch Statistics Adaptation
• Deep Adversarial Attention Alignment for Unsupervised Domain Adaptation: the Benefit of Target Expectation Maximization
• Adaptive Semantic Segmentation with a Strategic Curriculum of Proxy Labels
• Domain-Invariant Adversarial Learning for Unsupervised Domain Adaption
• Learning Pose-invariant 3D Object Reconstruction from Single-view Images
• ACE: Adapting to Changing Environments for Semantic Segmentation
• From source to target and back: symmetric bi-directional adaptive GAN
• Learning Factorized Representations for Open-set Domain Adaptation
• Invariant Representations from Adversarially Censored Autoencoders
• Distributional Adversarial Networks
• Label-Driven Reconstruction for Domain Adaptation in Semantic Segmentation
• Improving Unsupervised Domain Adaptation with Variational Information Bottleneck
• Domain Alignment with Triplets
• A Survey on Deep Domain Adaptation and Tiny Object Detection Challenges, Techniques and Datasets
• Adversarial Discriminative Sim-to-real Transfer of Visuo-motor Policies
• DART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
• MiniMax Entropy Network: Learning Category-Invariant Features for Domain Adaptation
• On Universal Black-Box Domain Adaptation
• Learning Cross-domain Generalizable Features by Representation Disentanglement
• GANerated Hands for Real-time 3D Hand Tracking from Monocular RGB
• Domain Attention Consistency for Multi-Source Domain Adaptation
• PM-GANs: Discriminative Representation Learning for Action Recognition Using Partial-modalities
• Be Like Water: Robustness to Extraneous Variables Via Adaptive Feature Normalization
• Learning from Multiple Cities: A Meta-Learning Approach for Spatial-Temporal Prediction
• Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation
• CDCGen: Cross-Domain Conditional Generation via Normalizing Flows and Adversarial Training
• Adversarial Attribute-Image Person Re-identification
• Beyond Domain Adaptation: Unseen Domain Encapsulation via Universal Non-volume Preserving Models
• Factorized Adversarial Networks for Unsupervised Domain Adaptation
• Learning-based Regularization for Cardiac Strain Analysis with Ability for Domain Adaptation
• Uncertainty-Aware Unsupervised Domain Adaptation in Object Detection
• Deep Domain Adaptation by Geodesic Distance Minimization
• Mind the Gap: Enlarging the Domain Gap in Open Set Domain Adaptation
• Machine learning with limited data
• Chameleon: Learning Model Initializations Across Tasks With Different Schemas
• CRAFT: Complementary Recommendations Using Adversarial Feature Transformer
• Learning not to Discriminate: Task Agnostic Learning for Improving Monolingual and Code-switched Speech Recognition
• Exploiting Local Feature Patterns for Unsupervised Domain Adaptation
• Structure-Attentioned Memory Network for Monocular Depth Estimation
• Unpaired Multi-Domain Image Generation via Regularized Conditional GANs
• Domain Adaptation and Image Classification via Deep Conditional Adaptation Network
• Dental pathology detection in 3D cone-beam CT
• Domain Adversarial Reinforcement Learning for Partial Domain Adaptation
• Zero-Annotation Object Detection with Web Knowledge Transfer
• Disentanglement for Discriminative Visual Recognition
• Identity Preserving Generative Adversarial Network for Cross-Domain Person Re-identification
• Zero-Shot Deep Domain Adaptation
• Unsupervised Domain Adaptation: from Simulation Engine to the RealWorld
• Towards Discriminative Representation Learning for Unsupervised Person Re-identification
• Cross-Domain Self-supervised Multi-task Feature Learning using Synthetic Imagery
• Improving Unsupervised Domain Adaptation by Reducing Bi-level Feature Redundancy
• Self-adaptive Re-weighted Adversarial Domain Adaptation
• Bias Reduction via End-to-End Shift Learning: Application to Citizen Science
• Contradictory Structure Learning for Semi-supervised Domain Adaptation
• Domain Adaptation Meets Disentangled Representation Learning and Style Transfer
• Mitigating the Impact of Speech Recognition Errors on Chatbot using Sequence-to-Sequence Model
• Encoding Robustness to Image Style via Adversarial Feature Perturbations
• Unsupervised Domain Adaptation for Learning Eye Gaze from a Million Synthetic Images: An Adversarial Approach
• Unsupervised Domain Adaptation using Generative Models and Self-ensembling
• Towards Accurate and Robust Domain Adaptation under Noisy Environments
• Domain Adaptation of Networks for Camera Pose Estimation: Learning Camera Pose Estimation Without Pose Labels
• Uncovering the Connections Between Adversarial Transferability and Knowledge Transferability
• Omni-GAN: On the Secrets of cGANs and Beyond
• Class Distribution Alignment for Adversarial Domain Adaptation
• Discriminative Active Learning for Domain Adaptation
• Domain Agnostic Real-Valued Specificity Prediction
• Exploring Textual and Speech information in Dialogue Act Classification with Speaker Domain Adaptation
• Multimodal Deep Domain Adaptation
• Cross-Dataset Adaptation for Visual Question Answering
• Generating Dialogue Responses from a Semantic Latent Space
• Stable Distribution Alignment Using the Dual of the Adversarial Distance
• Bi-Directional Generation for Unsupervised Domain Adaptation
• Learning Condensed and Aligned Features for Unsupervised Domain Adaptation Using Label Propagation
• Distributionally Robust Semi-Supervised Learning for People-Centric Sensing
• Regularized Wasserstein Means for Aligning Distributional Data
• AIM 2019 Challenge on Image Demoireing: Dataset and Study
• PortraitGAN for Flexible Portrait Manipulation
• Multichannel Semantic Segmentation with Unsupervised Domain Adaptation
• Towards Robust Pattern Recognition: A Review
• Open-Set Crowdsourcing using Multiple-Source Transfer Learning
• Instance-based Deep Transfer Learning
• Unsupervised Domain Expansion for Visual Categorization
• Semi-Supervised Adversarial Discriminative Domain Adaptation
• Personalized One-Shot Lipreading for an ALS Patient
• Self-Adaptive Partial Domain Adaptation
• Eliminating the Blind Spot: Adapting 3D Object Detection and Monocular Depth Estimation to 360° Panoramic Imagery
• Deep Discriminative Learning for Unsupervised Domain Adaptation
• Deep Cross-modality Adaptation via Semantics Preserving Adversarial Learning for Sketch-based 3D Shape Retrieval
• Domain Adaptation Regularization for Spectral Pruning
• Camera Invariant Feature Learning for Generalized Face Anti-spoofing
• How does the Combined Risk Affect the Performance of Unsupervised Domain Adaptation Approaches?
• Lethean Attack: An Online Data Poisoning Technique
• Person Re-identification with Bias-controlled Adversarial Training
• Deep Adversarial Domain Adaptation Based on Multi-layer Joint Kernelized Distance
• One Network for Multi-Domains: Domain Adaptive Hashing with Intersectant Generative Adversarial Network
• Task Guided Compositional Representation Learning for ZDA
• Escaping from Collapsing Modes in a Constrained Space
• Dynamic Adaptation on Non-Stationary Visual Domains
• T-SVDNet: Exploring High-Order Prototypical Correlations for Multi-Source Domain Adaptation
• Ensemble Multi-Source Domain Adaptation with Pseudolabels
• Leveraging Motion Priors in Videos for Improving Human Segmentation
• Exploring Data Aggregation and Transformations to Generalize across Visual Domains
• Parameter Reference Loss for Unsupervised Domain Adaptation
• Deep Domain Adaptation under Deep Label Scarcity
• CFEA: Collaborative Feature Ensembling Adaptation for Domain Adaptation in Unsupervised Optic Disc and Cup Segmentation
• Multiple Subspace Alignment Improves Domain Adaptation
• End-to-End Refinement Guided by Pre-trained Prototypical Classifier
• Efficient Video Understanding via Layered Multi Frame-Rate Analysis
• CrDoCo: Pixel-level Domain Transfer with Cross-Domain Consistency
• Adversarial Incremental Learning
• On-target Adaptation
• Domain Adaptive Monocular Depth Estimation With Semantic Information
• Image to Video Domain Adaptation Using Web Supervision
• Transfer Learning in Brain-Computer Interfaces with Adversarial Variational Autoencoders
• Artistic Object Recognition by Unsupervised Style Adaptation
• Generalization Challenges for Neural Architectures in Audio Source Separation
• Unsupervised Domain Adaptation for Cross-Subject Few-Shot Neurological Symptom Detection