Regularized Learning for Domain Adaptation under Label Shifts

Abstract

We propose Regularized Learning under Label shifts (RLLS), a principled and a practical domain-adaptation algorithm to correct for shifts in the label distribution between a source and a target domain. We first estimate importance weights using labeled source data and unlabeled target data, and then train a classifier on the weighted source samples. We derive a generalization bound for the classifier on the target domain which is independent of the (ambient) data dimensions, and instead only depends on the complexity of the function class. To the best of our knowledge, this is the first generalization bound for the label-shift problem where the labels in the target domain are not available. Based on this bound, we propose a regularized estimator for the small-sample regime which accounts for the uncertainty in the estimated weights. Experiments on the CIFAR-10 and MNIST datasets show that RLLS improves classification accuracy, especially in the low sample and large-shift regimes, compared to previous methods.

International Conference on Learning Representations (ICLR) 2019

Cited by in corpus (27)

• A review of domain adaptation without target labels
• WILDS: A Benchmark of in-the-Wild Distribution Shifts
• A Comprehensive Survey on Test-Time Adaptation under Distribution Shifts
• Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss
• What is the Effect of Importance Weighting in Deep Learning?
• Distribution Aligning Refinery of Pseudo-label for Imbalanced Semi-supervised Learning
• Towards Accurate Model Selection in Deep Unsupervised Domain Adaptation
• LEEP: A New Measure to Evaluate Transferability of Learned Representations
• Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift
• Domain Adaptation as a Problem of Inference on Graphical Models
• Disentangling Label Distribution for Long-tailed Visual Recognition
• A Review of Single-Source Deep Unsupervised Visual Domain Adaptation
• Transferable Calibration with Lower Bias and Variance in Domain Adaptation
• Minimax Lower Bounds for Transfer Learning with Linear and One-hidden Layer Neural Networks
• Domain adaptation under structural causal models
• Understanding the role of importance weighting for deep learning
• A Unified View of Label Shift Estimation
• Online Adaptation to Label Distribution Shift
• Did the Model Change? Efficiently Assessing Machine Learning API Shifts
• Coping with Label Shift via Distributionally Robust Optimisation
• Class-imbalanced Domain Adaptation: An Empirical Odyssey
• Sinkhorn Label Allocation: Semi-Supervised Classification via Annealed Self-Training
• Distribution-free uncertainty quantification for classification under label shift
• Aggregating From Multiple Target-Shifted Sources
• Mixture Proportion Estimation and PU Learning: A Modern Approach
• Weakly Supervised Learning Meets Ride-Sharing User Experience Enhancement
• Composed Fine-Tuning: Freezing Pre-Trained Denoising Autoencoders for Improved Generalization