While Unsupervised Domain Adaptation (UDA) algorithms, i.e., there are only labeled data from source domains, have been actively studied in recent years, most algorithms and theoretical results focus on Single-source Unsupervised Domain Adaptation (SUDA). However, in the practical scenario, labeled data can be typically collected from multiple diverse sources, and they might be different not only from the target domain but also from each other. Thus, domain adapters from multiple sources should not be modeled in the same way. Recent deep learning based Multi-source Unsupervised Domain Adaptation (MUDA) algorithms focus on extracting common domain-invariant representations for all domains by aligning distribution of all pairs of source and target domains in a common feature space. However, it is often very hard to extract the same domain-invariant representations for all domains in MUDA. In addition, these methods match distributions without considering domain-specific decision boundaries between classes. To solve these problems, we propose a new framework with two alignment stages for MUDA which not only respectively aligns the distributions of each pair of source and target domains in multiple specific feature spaces, but also aligns the outputs of classifiers by utilizing the domainspecific decision boundaries. Extensive experiments demonstrate that our method can achieve remarkable results on popular benchmark datasets for image classification.

No abstract

Mesenchymal stem cells, which are poorly immunogenic and have potent immunosuppressive activities, have emerged as promising cellular therapeutics for the treatment of several diseases. Mesenchymal-like cells derived from Wharton's Jelly, called umbilical cord matrix stem cells (UCMSCs), reportedly secrete a variety of cytokines and growth factors, acting as trophic suppliers. Here, we used UCMSCs to treat premature ovarian failure (POF). Ovarian function was evaluated by ovulation and the number of follicles. Apoptosis of the granulosa cells (GC) was analyzed by TUNEL staining. We found that after transplantation of the UCMSCs, apoptosis of cumulus cells in the ovarian damage model was reduced and the function of the ovary had been recovered. The sex hormone level was significantly elevated in mice treated with UCMSCs. The number of follicles in the treated group was higher than in the control group. Our results demonstrate that UCMSCs can effectively restore ovary functionality and reduce apoptosis of granulosa cells. We compared the RNA expression of the UCMSCs treated group with the POF model and wild-type control group and found that the UCMSC group is most similar to the wild-type group. Our experiments provide new information regarding the treatment of ovarian function failure.

Sequential recommendation, which aims to recommend next item that the user will likely interact in a near future, has become essential in various Internet applications. Existing methods usually consider the transition patterns between items, but ignore the transition patterns between features of items. We argue that only the item-level sequences cannot reveal the full sequential patterns, while explicit and implicit feature-level sequences can help extract the full sequential patterns. In this paper, we propose a novel method named Feature-level Deeper Self-Attention Network (FDSA) for sequential recommendation. Specifically, FDSA first integrates various heterogeneous features of items into feature sequences with different weights through a vanilla mechanism. After that, FDSA applies separated self-attention blocks on item-level sequences and feature-level sequences, respectively, to model item transition patterns and feature transition patterns. Then, we integrate the outputs of these two blocks to a fully-connected layer for next item recommendation. Finally, comprehensive experimental results demonstrate that considering the transition relationships between features can significantly improve the performance of sequential recommendation.

No abstract

No abstract

The amount of erythrocyte-derived microRNAs (miRNAs) represents the majority of miRNAs expressed in whole blood. miR-451, miR-144, and miR-486, which are abundant in red blood cells (RBCs), are involved in the process of erythropoiesis and disease occurrence. Moreover, erythrocyte-derived miRNAs have been reported to be potential biomarkers of specific diseases. However, the function and underlying mechanisms of miRNAs derived from erythrocytes remain unclear. Based on a review of previously published literature, we discuss several possible pathways by which RBC miRNAs may function and propose that RBCs may serve as repositories of miRNAs in the circulatory system and participate in the regulation of gene expression mainly via the transfer of miRNAs from erythrocyte extracellular vesicles (EVs). In the whole blood, there are still other important cell types such as leukocytes and platelets harboring functional miRNAs, and hemolysis also exists, which limit the abundance of miRNAs as disease biomarkers, and thus, miRNA studies on RBCs may be impacted. In the future, the role of RBCs in the regulation of normal physiological functions of the body and the entire circulatory system under pathological states, if any, remains to be determined.

No abstract