Unsupervised Domain Adaptation by Backpropagation

Abstract

Top-performing deep architectures are trained on massive amounts of labeled data. In the absence of labeled data for a certain task, domain adaptation often provides an attractive option given that labeled data of similar nature but from a different domain (e.g. synthetic images) are available. Here, we propose a new approach to domain adaptation in deep architectures that can be trained on large amount of labeled data from the source domain and large amount of unlabeled data from the target domain (no labeled target-domain data is necessary). As the training progresses, the approach promotes the emergence of "deep" features that are (i) discriminative for the main learning task on the source domain and (ii) invariant with respect to the shift between the domains. We show that this adaptation behaviour can be achieved in almost any feed-forward model by augmenting it with few standard layers and a simple new gradient reversal layer. The resulting augmented architecture can be trained using standard backpropagation. Overall, the approach can be implemented with little effort using any of the deep-learning packages. The method performs very well in a series of image classification experiments, achieving adaptation effect in the presence of big domain shifts and outperforming previous state-of-the-art on Office datasets.

References in corpus (2)

• Caffe: Convolutional Architecture for Fast Feature Embedding
• Deep Domain Confusion: Maximizing for Domain Invariance

Cited by in corpus (92)

• Deep Subdomain Adaptation Network for Image Classification
• FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• Adversarial Discriminative Domain Adaptation
• Aligning Domain-specific Distribution and Classifier for Cross-domain Classification from Multiple Sources
• Maximum Density Divergence for Domain Adaptation
• Multi-Representation Adaptation Network for Cross-domain Image Classification
• Weakly Supervised Adversarial Domain Adaptation for Semantic Segmentation in Urban Scenes
• Deep Transfer Learning for Person Re-identification
• Collaborative Unsupervised Domain Adaptation for Medical Image Diagnosis
• A Survey of mmWave-based Human Sensing: Technology, Platform and Applications
• Deep Residual Correction Network for Partial Domain Adaptation
• Silent Speech Interfaces for Speech Restoration: A Review
• DGCN: Diversified Recommendation with Graph Convolutional Networks
• Learning Disentangled Semantic Representation for Domain Adaptation
• Multi-Source Unsupervised Domain Adaptation via Pseudo Target Domain
• Cross-regional oil palm tree counting and detection via multi-level attention domain adaptation network
• Missing-Class-Robust Domain Adaptation by Unilateral Alignment for Fault Diagnosis
• Generalized Domain Conditioned Adaptation Network
• Meta Balanced Network for Fair Face Recognition
• Adversarial Speaker Verification
• Multi-modal Deep Analysis for Multimedia
• Uncertainty Aware Learning for High Energy Physics
• Cross-Market Product Recommendation
• Learning Target Domain Specific Classifier for Partial Domain Adaptation
• Speaker Adaptation for Attention-Based End-to-End Speech Recognition
• Adversarial Speaker Adaptation
• MT-UDA: Towards Unsupervised Cross-modality Medical Image Segmentation with Limited Source Labels
• Domain Adaptive Person Re-Identification via Coupling Optimization
• Robust Domain Misinformation Detection via Multi-modal Feature Alignment
• Inferring Latent Domains for Unsupervised Deep Domain Adaptation
• A Noise-Robust Self-supervised Pre-training Model Based Speech Representation Learning for Automatic Speech Recognition
• An Unsupervised Domain Adaptation Model based on Dual-module Adversarial Training
• Multi-source Domain Adaptation for Text-independent Forensic Speaker Recognition
• Image to Image Translation for Domain Adaptation
• Knowledge Distillation Methods for Efficient Unsupervised Adaptation Across Multiple Domains
• DVR: Micro-Video Recommendation Optimizing Watch-Time-Gain under Duration Bias
• Continuous Gesture Recognition from sEMG Sensor Data with Recurrent Neural Networks and Adversarial Domain Adaptation
• Adaptive Betweenness Clustering for Semi-Supervised Domain Adaptation
• All about Structure: Adapting Structural Information across Domains for Boosting Semantic Segmentation
• Coupling Adversarial Learning with Selective Voting Strategy for Distribution Alignment in Partial Domain Adaptation
• Cross-Domain Aspect Extraction using Transformers Augmented with Knowledge Graphs
• A Dataset and Benchmark Towards Multi-Modal Face Anti-Spoofing Under Surveillance Scenarios
• DUSA: Decoupled Unsupervised Sim2Real Adaptation for Vehicle-to-Everything Collaborative Perception
• Prototypical Partial Optimal Transport for Universal Domain Adaptation
• Semi-Supervised Disentangled Framework for Transferable Named Entity Recognition
• A Flexible Selection Scheme for Minimum-Effort Transfer Learning
• Disentangled speaker and nuisance attribute embedding for robust speaker verification
• Sensor-invariant Fingerprint ROI Segmentation Using Recurrent Adversarial Learning
• Improving Domain Adaptation Through Class Aware Frequency Transformation
• ACE: Adapting to Changing Environments for Semantic Segmentation
• Bridged-GNN: Knowledge Bridge Learning for Effective Knowledge Transfer
• Learning Common and Specific Features for RGB-D Semantic Segmentation with Deconvolutional Networks
• EllSeg-Gen, towards Domain Generalization for head-mounted eyetracking
• Improving Automated Sonar Video Analysis to Notify About Jellyfish Blooms
• A Learnable Self-supervised Task for Unsupervised Domain Adaptation on Point Clouds
• Attentive Adversarial Learning for Domain-Invariant Training
• Domain Alignment with Triplets
• VisDA-2021 Competition Universal Domain Adaptation to Improve Performance on Out-of-Distribution Data
• Transfering Low-Frequency Features for Domain Adaptation
• Micro-video Tagging via Jointly Modeling Social Influence and Tag Relation
• A Multi Camera Unsupervised Domain Adaptation Pipeline for Object Detection in Cultural Sites through Adversarial Learning and Self-Training
• Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation
• Predictive Uncertainty-based Bias Mitigation in Ranking
• An Adversarial Domain Separation Framework for Septic Shock Early Prediction Across EHR Systems
• Alioth: A Machine Learning Based Interference-Aware Performance Monitor for Multi-Tenancy Applications in Public Cloud
• Emo-DNA: Emotion Decoupling and Alignment Learning for Cross-Corpus Speech Emotion Recognition
• DADNN: Multi-Scene CTR Prediction via Domain-Aware Deep Neural Network
• Small Towers Make Big Differences
• Domain Adversarial Reinforcement Learning
• CrowdTransfer: Enabling Crowd Knowledge Transfer in AIoT Community
• BiPC: Bidirectional Probability Calibration for Unsupervised Domain Adaption
• A Theory of Output-Side Unsupervised Domain Adaptation
• DeMIAN: Deep Modality Invariant Adversarial Network
• Cross-Domain Self-supervised Multi-task Feature Learning using Synthetic Imagery
• Union-set Multi-source Model Adaptation for Semantic Segmentation
• Adapting a FrameNet Semantic Parser for Spoken Language Understanding Using Adversarial Learning
• An Attribute-Aligned Strategy for Learning Speech Representation
• Self-balanced Learning For Domain Generalization
• OmniPrint: A Configurable Printed Character Synthesizer
• Transfer Learning for Aided Target Recognition: Comparing Deep Learning to other Machine Learning Approaches
• I3DOL: Incremental 3D Object Learning without Catastrophic Forgetting
• The Unreasonable Effectiveness of Large Language-Vision Models for Source-free Video Domain Adaptation
• Unsupervised Multi-Domain Multimodal Image-to-Image Translation with Explicit Domain-Constrained Disentanglement
• G2R Bound: A Generalization Bound for Supervised Learning from GAN-Synthetic Data
• Multichannel Semantic Segmentation with Unsupervised Domain Adaptation
• Dual Domain-Adversarial Learning for Audio-Visual Saliency Prediction
• Synthesizing Photorealistic Images with Deep Generative Learning
• Test-Time Adaptation with Principal Component Analysis
• The Role of Embedding Complexity in Domain-invariant Representations
• Learning Deep Features for Shape Correspondence with Domain Invariance
• Parameter Reference Loss for Unsupervised Domain Adaptation