Butterfly: One-step Approach towards Wildly Unsupervised Domain Adaptation

Abstract

In unsupervised domain adaptation (UDA), classifiers for the target domain (TD) are trained with clean labeled data from the source domain (SD) and unlabeled data from TD. However, in the wild, it is difficult to acquire a large amount of perfectly clean labeled data in SD given limited budget. Hence, we consider a new, more realistic and more challenging problem setting, where classifiers have to be trained with noisy labeled data from SD and unlabeled data from TD -- we name it wildly UDA (WUDA). We show that WUDA ruins all UDA methods if taking no care of label noise in SD, and to this end, we propose a Butterfly framework, a powerful and efficient solution to WUDA. Butterfly maintains four deep networks simultaneously, where two take care of all adaptations (i.e., noisy-to-clean, labeled-to-unlabeled, and SD-to-TD-distributional) and then the other two can focus on classification in TD. As a consequence, Butterfly possesses all the conceptually necessary components for solving WUDA. Experiments demonstrate that, under WUDA, Butterfly significantly outperforms existing baseline methods.

arXiv commentPrevious version of this paper has been accepted by the NeurIPS2019 Workshop on Learning Transferable Skills (https://www.skillsworkshop.ai/schedule.html)

References in corpus (15)

• Learning Transferable Features with Deep Adaptation Networks
• Classification with Noisy Labels by Importance Reweighting
• Domain Adaptation: Learning Bounds and Algorithms
• Asymmetric Tri-training for Unsupervised Domain Adaptation
• A DIRT-T Approach to Unsupervised Domain Adaptation
• A Closer Look at Memorization in Deep Networks
• Bridging Theory and Algorithm for Domain Adaptation
• Learning with Augmented Features for Heterogeneous Domain Adaptation
• Positive-Unlabeled Learning with Non-Negative Risk Estimator
• Learning Deep Kernels for Non-Parametric Two-Sample Tests
• Minimax Statistical Learning with Wasserstein Distances
• A Testbed for Cross-Dataset Analysis
• Cross Language Text Classification via Subspace Co-Regularized Multi-View Learning
• Transfer Learning with Label Noise
• Causal Generative Domain Adaptation Networks

Cited by in corpus (4)

• A Survey of Label-noise Representation Learning: Past, Present and Future
• Autonomous Cross Domain Adaptation under Extreme Label Scarcity
• Domain Adaptation for Learning Generator from Paired Few-Shot Data
• How does the Combined Risk Affect the Performance of Unsupervised Domain Adaptation Approaches?