Multi-Source Domain Adaptation with Mixture of Experts

Abstract

We propose a mixture-of-experts approach for unsupervised domain adaptation from multiple sources. The key idea is to explicitly capture the relationship between a target example and different source domains. This relationship, expressed by a point-to-set metric, determines how to combine predictors trained on various domains. The metric is learned in an unsupervised fashion using meta-training. Experimental results on sentiment analysis and part-of-speech tagging demonstrate that our approach consistently outperforms multiple baselines and can robustly handle negative transfer.

arXiv comment11 pages, EMNLP 2018

References in corpus (8)

• Prototypical Networks for Few-shot Learning
• Deep Domain Confusion: Maximizing for Domain Invariance
• Matching Networks for One Shot Learning
• Domain Separation Networks
• Wasserstein Distance Guided Representation Learning for Domain Adaptation
• Knowledge Adaptation: Teaching to Adapt
• Multinomial Adversarial Networks for Multi-Domain Text Classification
• Parsing Tweets into Universal Dependencies

Cited by in corpus (3)

• Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift
• NeuraCrypt: Hiding Private Health Data via Random Neural Networks for Public Training
• Universal Multi-Source Domain Adaptation