A Theory of Multiple-Source Adaptation with Limited Target Labeled Data
This work addresses domain adaptation challenges for machine learning practitioners when target labels are scarce, though it appears incremental as it builds on existing model selection ideas.
The paper tackles the multiple-source domain adaptation problem with limited labeled target data, proposing a new family of algorithms based on model selection that achieves favorable theoretical guarantees and demonstrates practical effectiveness in experiments.
We present a theoretical and algorithmic study of the multiple-source domain adaptation problem in the common scenario where the learner has access only to a limited amount of labeled target data, but where the learner has at disposal a large amount of labeled data from multiple source domains. We show that a new family of algorithms based on model selection ideas benefits from very favorable guarantees in this scenario and discuss some theoretical obstacles affecting some alternative techniques. We also report the results of several experiments with our algorithms that demonstrate their practical effectiveness.