In the following study, an innovative domain adaptation technique is proposed. dCor based Domain Adaptation or dDA technique is based on maximum mean discrepancy (MMD) and distance correlation (dCor); a powerful yet general correlation measure which is applicable to arbitrary-dimensional random variables. By projecting the samples to a common latent feature space, dDA minimizes the discrepancy between the source and target distributions while preserving the structural information of the data. The proposed dDA algorithm can be easily implemented and it has a closed-form and simple solution. Extensive analyses across various real-world and sentiment analysis benchmark data sets indicate that our algorithm is the method of choice; as it offers superior results in comparison with several state-of-the-art domain adaptation approaches in the literature in both unsupervised and semi-supervised settings.

在下面的研究中，提出了一种创新的领域适应技术。基于dCor的域自适应或dDA技术基于最大平均差异（MMD）和距离相关性（dCor）；一种强大而通用的相关度量，适用于任意维随机变量。通过将样本投影到一个共同的潜在特征空间，dDA可以最大程度地减少源分布和目标分布之间的差异，同时保留数据的结构信息。所提出的dDA算法可以轻松实现，并且具有封闭形式和简单的解决方案。跨各种现实世界和情感分析基准数据集的大量分析表明，我们的算法是首选方法。