Institute of Machine Intelligence (IMI), University of Shanghai for Science and Technology, Shanghai 200093, China; Technical Aspects of Multimodal Systems (TAMS) Group, Universität Hamburg, Hamburg D-22527, Germany; University of Electronic Science and Technology of China, Chengdu 611731, China.
Institute of Machine Intelligence (IMI), University of Shanghai for Science and Technology, Shanghai 200093, China.
Neural Netw. 2022 Aug;152:467-478. doi: 10.1016/j.neunet.2022.05.015. Epub 2022 May 21.
Recently, source data-free unsupervised domain adaptation (SFUDA) attracts increasing attention. Current work shows that the geometry of the target data is helpful to solving this challenging problem. However, these methods define the geometric structures in Euclidean space. The geometry cannot completely draw the semantic relationship between the target data distributed on a manifold. This article proposed a new SFUDA method, semantic consistency learning on manifold (SCLM), to address this problem. Firstly, we generated pseudo-labels for the target data using a new clustering method, EntMomClustering, that enhanced k-means clustering by fusing the entropy momentum. Secondly, we constructed semantic neighbor topology (SNT) to capture complete geometric information on the manifold. Specifically, in SNT, the global neighbor was detected by a developed collaborative representation-based manifold projection, while the local neighbors were obtained by similarity comparison. Thirdly, we performed a semantic consistency learning on SNT to drive a new kind of deep clustering where SNT was taken as the basic clustering unit. To ensure SNT move as entirety, in the developed objective, the entropy regulator was constructed based on a semantic mixture fused on SNT, while the self-supervised regulator encouraged similar classification on SNT. Experiments on three benchmark datasets show that our method achieves state-of-the-art results. The code is available on https://github.com/tntek/SCLM.
最近，基于源数据的无监督领域自适应（Source Data-Free Unsupervised Domain Adaptation，SFUDA）受到了越来越多的关注。现有研究表明目标数据的几何结构有助于解决这一具有挑战性的问题。然而，这些方法在欧几里得空间中定义了几何结构。这种几何结构不能完全描绘流形上分布的目标数据的语义关系。本文提出了一种新的 SFUDA 方法，即流形上的语义一致性学习（Semantic Consistency Learning on Manifold，SCLM），以解决这个问题。首先，我们使用一种新的聚类方法——EntMomClustering 为目标数据生成伪标签，该方法通过融合熵动量增强了 k-means 聚类。其次，我们构建了语义邻域拓扑（Semantic Neighbor Topology，SNT）来捕捉流形上完整的几何信息。具体来说，在 SNT 中，全局邻居是通过一种基于协同表示的流形投影算法检测到的，而局部邻居是通过相似度比较得到的。然后，我们在 SNT 上进行语义一致性学习，以驱动一种新的深度聚类，其中 SNT 被用作基本聚类单元。为了确保 SNT 整体移动，在开发的目标函数中，基于 SNT 上融合的语义混合，构建了熵正则化项，而自监督正则化项则鼓励 SNT 上的相似分类。在三个基准数据集上的实验表明，我们的方法取得了最先进的结果。代码可在 https://github.com/tntek/SCLM 上获取。
