1: \begin{abstract}

2: In leveraging manifold learning in domain adaptation \textbf{(DA)}, graph embedding-based \textbf{DA} methods have shown their effectiveness in preserving data manifold through the Laplace graph. However, current graph embedding \textbf{DA} methods suffer from two issues: 1). they are only concerned with preservation of the underlying data structures in the embedding and ignore sub-domain adaptation, which requires taking into account intra-class similarity and inter-class dissimilarity, thereby leading to negative transfer; 2). manifold learning is proposed across different feature/label spaces separately, thereby hindering unified comprehensive manifold learning. In this paper, starting from our previous \textbf{DGA-DA}, we propose a novel \textbf{DA} method, namely \textit{A}ttention \textit{R}egularized Laplace \textit{G}raph-based \textit{D}omain \textit{A}daptation (\textbf{ARG-DA}), to remedy the aforementioned issues. Specifically, by weighting the importance across different sub-domain adaptation tasks, we propose the \textit{A}ttention \textit{R}egularized Laplace \textbf{G}raph for class aware \textbf{DA}, thereby generating the attention regularized \textbf{DA}. 	Furthermore, using a specifically designed \textbf{FEEL} strategy, our approach dynamically unifies alignment of the manifold structures across different feature/label spaces, thus leading to comprehensive manifold learning. Comprehensive experiments are carried out to verify the effectiveness of the proposed \textbf{DA} method, which consistently outperforms the state of the art \textbf{DA} methods on 7 standard \textbf{DA} benchmarks, \textit{i.e.}, 37 cross-domain image classification tasks including object, face, and digit images.  An in-depth analysis of the proposed \textbf{DA} method is also discussed, including sensitivity, convergence, and robustness.

3: \end{abstract}

4: