Improving imbalance classification via ensemble learning based on two-stage learning.

The excellent performance of deep neural networks on image classification tasks depends on a large-scale high-quality dataset. However, the datasets collected from the real world are typically biased in their distribution, which will lead to a sharp decline in model performance, mainly because an imbalanced distribution results in the prior shift and covariate shift. Recent studies have typically used a two-stage learning method consisting of two rebalancing strategies to solve these problems, but the combination of partial rebalancing strategies will damage the representational ability of the networks. In addition, the two-stage learning method is of little help in addressing the problem of covariate shift. To solve the above two issues, we first propose a sample logit-aware reweighting method called (SLA), which can not only repair the weights of majority class hard samples and minority class samples but will also integrate with logit adjustment to form a stable two-stage learning strategy. Second, to solve the covariate shift problem, inspired by ensemble learning, we propose a multi-domain expert specialization model, which can achieve a more comprehensive decision by averaging expert classification results from multiple different domains. Finally, we combine SLA and logit adjustment into a two-stage learning method and apply our model to the CIFAR-LT and ImageNet-LT datasets. Compared with the most advanced methods, our experimental results show excellent performance.

DNM3OS Enhances the Apoptosis and Senescence of Spermatogonia Associated with Nonobstructive Azoospermia by Providing miR-214-5p and Decreasing E2F2 Expression.

Background: Nonobstructive azoospermia (NOA) is a complex disease characterized by the spermatogenic dysfunction of testicular tissues. The roles played by long noncoding RNAs (lncRNAs) in NOA pathogenesis have not been extensively studied. Methods: Microarray assays were performed on samples of testicular biopsy tissue obtained from patients with NOA for the purpose of identifying differentially expressed lncRNAs and messenger RNA (mRNA) transcripts, and the results were verified by quantitative real-time polymerase chain reaction. Mouse-derived GC-1 spermatogonia (spg) cells undergoing treatment with Adriamycin (ADR) were used to investigate the biological functions of the selected lncRNAs in vitro. The target microRNAs (miRNAs) of lncRNAs and the target mRNAs of miRNAs were predicted by a bioinformatics analysis. Functional studies performed using the CCK-8 assay, EdU incorporation assay, apoptosis detection, and senescence-associated ß-galactosidase (SA-ß-Gal) staining were conducted using GC-1 spg cells. Results: Totals of 2,652 lncRNAs and 2,625 mRNAs were found to be differentially expressed in the testicular tissue of NOA patients when compared with patients in a control group. Dynamin 3 opposite strand (DNM3OS) was a provider of pe-miR-214-5p that positively regulates miR-214-5p expression in GC-1 spg cells. The E2 factor (E2F) family of transcription factor 2 (E2F2) was initially predicted and subsequently verified to be a downstream gene of miR-214-5p. E2F2 expression was upregulated after DNM3OS knockdown in ADR-treated GC-1 spg cells. Moreover, knockdown of either DNM3OS or miR-214-5p significantly alleviated ADR-induced decreases in cellular activity and proliferation, as well as increases in apoptosis and senescence of mouse spermatogonial GC-1 spg cells. Conclusions: DNM3OS was found to regulate the apoptosis and senescence of spermatogonia by providing miR-214-5p and decreasing E2F2 expression, suggesting it as a novel target for gene therapy of male infertility.

A modified method by differential adhesion for enrichment of bladder cancer stem cells

ABSTRACT Purpose: In a previous study the vaccine was effective against bladder cancer in a mouse model. However, a small portion of tumors regrew because the vaccine could not eliminate bladder cancer stem cells (CSCs). In this study, we showed a modified method for the isolation of bladder CSCs using a combination of differential adhesion method and serum-free culture medium (SFM) method. Materials and Methods: Trypsin-resistant cells and trypsin-sensitive cells were isolated from MB49, EJ and 5637 cells by a combination of differential adhesion method and SFM method. The CSCs characterizations of trypsin-resistant cells were verified by the flow cytometry, the western blotting, the quantitative polymerase chain reaction, the resistance to chemotherapy assay, the transwell assay, and the tumor xenograft formation assay. Results: Trypsin-resistant cells were isolated and identified in CSCs characters, with high expression of CSCs markers, higher resistance to chemotherapy, greater migration in vitro, and stronger tumorigenicity in vivo. Conclusion: Trypsin-resistant cells displayed specific CSCs properties. Our study showed trypsin-resistant cells were isolated successfully with a modified method using a combination of differential adhesion method and SFM method.