Learning Sampling Policies for Domain Adaptation
This addresses domain adaptation for visual classification, but appears incremental as it builds on existing methods with a specific sampling approach.
The paper tackles semi-supervised domain adaptation for classification by using deep Q-learning to sample target domain data with noisy labels from a source network, aiming to maximize accuracy on a small annotated target subset. Experiments show the learned policies improve visual classifier accuracies over baselines.
We address the problem of semi-supervised domain adaptation of classification algorithms through deep Q-learning. The core idea is to consider the predictions of a source domain network on target domain data as noisy labels, and learn a policy to sample from this data so as to maximize classification accuracy on a small annotated reward partition of the target domain. Our experiments show that learned sampling policies construct labeled sets that improve accuracies of visual classifiers over baselines.