Domain-Adversarial Neural Networks

2014/12/15 by Hana Ajakan, Pascal Germain, Ajakan, Hana +7 · 38 citations
Computer Science · Mathematics · #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Neural and Evolutionary Computing (cs.NE) #cs.LG #cs.NE #stat.ML

paper · pdf · doi:10.48550/arxiv.1412.4446

The first version of this paper was accepted at the "Second Workshop on Transfer and Multi-Task Learning: Theory meets Practice" (NIPS 2014, Montreal, Canada). See: https://sites.google.com/site/multitaskwsnips2014/

openalex publication_date 2014/12/15 · arxiv created 2015/02/09 · arxiv updated 2015/02/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We introduce a new representation learning algorithm suited to the context of domain adaptation, in which data at training and test time come from similar but different distributions. Our algorithm is directly inspired by theory on domain adaptation suggesting that, for effective domain transfer to be achieved, predictions must be made based on a data representation that cannot discriminate between the training (source) and test (target) domains. We propose a training objective that implements this idea in the context of a neural network, whose hidden layer is trained to be predictive of the classification task, but uninformative as to the domain of the input. Our experiments on a sentiment analysis classification benchmark, where the target domain data available at training time is unlabeled, show that our neural network for domain adaption algorithm has better performance than either a standard neural network or an SVM, even if trained on input features extracted with the state-of-the-art marginalized stacked denoising autoencoders of Chen et al. (2012).

Citations

• Domain Adaptation: Learning Bounds and Algorithms

Cited by

• Learning Depth from Monocular Videos Using Synthetic Data: A Temporally-Consistent Domain Adaptation Approach
• Improving Artifact Robustness for CT Deep Learning Models Without Labeled Artifact Images via Domain Adaptation
• Multi-Level Heterogeneous Knowledge Transfer Network on Forward Scattering Center Model for Limited Samples SAR ATR
• Learning Purified Feature Representations from Task-irrelevant Labels
• CoDoL: Conditional Domain Prompt Learning for Out-of-Distribution Generalization
• Learning Transferable Features with Deep Adaptation Networks
• Invariant Risk Minimization Games
• Domain Separation Networks
• CORE-ReID V2: Advancing the Domain Adaptation for Object Re-Identification with Optimized Training and Ensemble Fusion
• Metric-Learning-Assisted Domain Adaptation
• Synthetic-to-Real Camouflaged Object Detection
• Putting Fairness Principles into Practice: Challenges, Metrics, and Improvements
• Empirical or Invariant Risk Minimization? A Sample Complexity Perspective
• DAMIA: Leveraging Domain Adaptation as a Defense against Membership Inference Attacks
• Domain-Adversarial Multi-Task Framework for Novel Therapeutic Property Prediction of Compounds
• Ancient Painting to Natural Image: A New Solution for Painting Processing
• Dealing with Nuisance Parameters using Machine Learning in High Energy Physics: a Review
• Online Monotone Optimization
• Domain Adaptation for Real-World Single View 3D Reconstruction
• Learning Adversarially Fair and Transferable Representations
• Knowledge Projection for Deep Neural Networks
• Geometry-Aware Symmetric Domain Adaptation for Monocular Depth Estimation
• Transfer learning for Remaining Useful Life Prediction Based on Consensus Self-Organizing Models
• Adversarial Alignment of Class Prediction Uncertainties for Domain Adaptation
• AD-Cluster: Augmented Discriminative Clustering for Domain Adaptive Person Re-identification
• Beyond the Deep Metric Learning: Enhance the Cross-Modal Matching with Adversarial Discriminative Domain Regularization
• Human Heterogeneity Invariant Stress Sensing
• Unsupervised Domain Adaptive Learning via Synthetic Data for Person Re-identification
• When Shift Happens - Confounding Is to Blame
• X-MethaneWet: A Cross-scale Global Wetland Methane Emission Benchmark Dataset for Advancing Science Discovery with AI
• Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation
• Similarity-preserving Image-image Domain Adaptation for Person Re-identification
• Adversarial Multi-task Learning for Text Classification
• Rethinking Distributional Matching Based Domain Adaptation
• Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks
• Unsupervised transfer learning for anomaly detection: Application to complementary operating condition transfer
• Identity Preserving Generative Adversarial Network for Cross-Domain Person Re-identification
• Learning Weighted Representations for Generalization Across Designs

Related

• Domain-Adversarial Training of Neural Networks
• Multiple Source Domain Adaptation with Adversarial Training of Neural Networks
• Domain-Symmetric Networks for Adversarial Domain Adaptation
• What's in a Domain? Learning Domain-Robust Text Representations using Adversarial Training
• Domain-Invariant Adversarial Learning for Unsupervised Domain Adaption