Bridging Theory and Algorithm for Domain Adaptation
This work addresses the problem of improving domain adaptation methods for machine learning practitioners by bridging theoretical insights with practical algorithms, though it is incremental as it builds on existing theories.
The paper tackles the gap between theory and algorithm in unsupervised domain adaptation by introducing Margin Disparity Discrepancy, a novel measurement with generalization bounds, and shows that the resulting algorithm achieves state-of-the-art accuracies on challenging tasks.
This paper addresses the problem of unsupervised domain adaption from theoretical and algorithmic perspectives. Existing domain adaptation theories naturally imply minimax optimization algorithms, which connect well with the domain adaptation methods based on adversarial learning. However, several disconnections still exist and form the gap between theory and algorithm. We extend previous theories (Mansour et al., 2009c; Ben-David et al., 2010) to multiclass classification in domain adaptation, where classifiers based on the scoring functions and margin loss are standard choices in algorithm design. We introduce Margin Disparity Discrepancy, a novel measurement with rigorous generalization bounds, tailored to the distribution comparison with the asymmetric margin loss, and to the minimax optimization for easier training. Our theory can be seamlessly transformed into an adversarial learning algorithm for domain adaptation, successfully bridging the gap between theory and algorithm. A series of empirical studies show that our algorithm achieves the state of the art accuracies on challenging domain adaptation tasks.