Deep neural networks often require copious amount of labeled-data to train
their scads of parameters. Training larger and deeper networks is hard without
appropriate regularization, particularly while using a small dataset.
Laterally, collecting well-annotated data is expensive, time-consuming and
often infeasible. A popular way to regularize these networks is to simply train
the network with more data from an alternate representative dataset. This can
lead to adverse effects if the statistics of the representative dataset are
dissimilar to our target. This predicament is due to the problem of domain
shift. Data from a shifted domain might not produce bespoke features when a
feature extractor from the representative domain is used. In this paper, we
propose a new technique ($d$-SNE) of domain adaptation that cleverly uses
stochastic neighborhood embedding techniques and a novel modified-Hausdorff
distance. The proposed technique is learnable end-to-end and is therefore,
ideally suited to train neural networks. Extensive experiments demonstrate that
$d$-SNE outperforms the current states-of-the-art and is robust to the
variances in different datasets, even in the one-shot and semi-supervised
learning settings. $d$-SNE also demonstrates the ability to generalize to
multiple domains concurrently.

该论文提出了一种名为 $d$-SNE 的域适应技术，通过使用随机邻域嵌入技术和一种新的修改 Hausdorff 距离，学习端到端，使其非常适合训练神经网络，并且在实验中表现出优于当前最先进技术的性能，并且能够同时推广到多个领域。