Search

In vision domain, large-scale natural datasets typically exhibit long-tailed distribution which has large class imbalance between head and tail classes. This distribution poses difficulty in learning good representations for tail classes. Recent developments have shown good long-tailed model can be learnt by decoupling the training into representation learning and classifier balancing. However, these works pay insufficient consideration on the long-tailed effect on representation learning. In this work, we propose interpolative centroid contrastive learning (ICCL) to improve long-tailed representation learning. ICCL interpolates two images from a class-agnostic sampler and a class-aware sampler, and trains the model such that the representation of the interpolative image can be used to retrieve the centroids for both source classes. We demonstrate the effectiveness of our approach on multiple long-tailed image classification benchmarks. Our result shows a significant accuracy gain of 2.8% on the iNaturalist 2018 dataset with a real-world long-tailed distribution., Comment: Add in acknowledgment

Junior high school students are at a critical stage of adolescent development. These children are emotionally sensitive and prone to mental health problems. A survey of the mental health status of junior high school pupils found that the detection rate of learning anxiety was highest in junior high school pupils and increased with grade level. Therefore, it is necessary to investigate the causes and influencing factors of learning anxiety in depth. This study analyzes and compares the existing literature on the effects of interpersonal relationships on learning anxiety in junior high school students. Results show that normal parent-child relationships, supportive teacher-student relationships, and peer support were all negatively associated with learning anxiety. In contrast, unhealthy interpersonal relationships will increase students learning anxiety level. This paper explores how parent-child, teacher-student, and peer relationships influence learning anxiety among junior high school students, reviewing and collating data to derive both a dominant-subordinate and an integrative perspective. Overall, interpersonal relationships affect learning anxiety in junior high students in different ways and to various degrees, and interpersonal relationships affect the generation and mitigation of learning anxiety in junior high students to some extent.

Carrying out gender equality education in junior middle schools is one of the significant objectives and tasks in the Development of Chinese Women (2021-2030). The adolescent stage is a crucial period for gender equality education. It is significant for teenagers to implement and promote gender equality education in junior high school. This paper starts with the current situation of gender equality education in junior middle schools in our country. It analyzes the existing problems of gender education from four aspects: school education policy, curriculum setting, teachers' gender education literacy, and classroom teaching. Based on the problem analysis, this paper concludes that schools and teachers should start by establishing gender equality consciousness, starting from campus activities, curriculum system design, and classroom teaching activities, and put forward corresponding suggestions and countermeasures.

Inflammation and immune dysregulation are involved in the pathogenesis of diabetic macular edema (DME). The progressive increase of neutrophils in peripheral blood can lead to the increase of the number of neutrophils in the retina, thus leading to the sustained damage of the retinal vascular system and the destruction of the blood retinal barrier (BRB); lymphocytes play a protective role in vascular diseases caused by type 2 diabetes mellitus (T2DM). The purpose of this study was to study the relationship between the changes of leukocytes and their classification in peripheral blood and the occurrence and progression of DME in patients with T2DM.A retrospective analysis was made on 81 patients with T2DM with DME (DME group) hospitalized in our hospital from January 2019 to December 2020. According to the morphological characteristics of macular edema in optical coherence tomography (OCT), they were divided into early DME group (n=33) and late DME group (n=48); 33 patients with diabetes retinopathy (DR) but without DME matched in age and course of disease served as the control group (NO-DME group). The clinical parameters assessed included eye examination, OCT results, WBCs and subtypes, blood glucose, and glycosylated hemoglobin.Compared with NO-DME group (n=33), Neutrophils% in DME group (n=81) was higher (57.37±9.52 vs. 63.27±7.85; P=0.001); Monocyte% (7.63±1.77 vs. 6.88±1.83; P=0.047) and lymphocyte% (30.35±9.51 vs. 27.26±6.59; P=0.032) were decreased. The optimal model was obtained with R 4.0.5 software. With other relevant variables being the same, females had a significantly increased risk of DME (b=1.273, P=0.015), %neutrophils was significantly associated with increased risk of DME (b=0.152, P=0.0006), and %lymphocytes was significantly associated with a reduced risk of DME (b=-0.027, P=0.179). However, in the early and late DME groups, no significant differences in biological markers were found, and a high-quality model was not obtained.In this preliminary study, %neutrophils is associated with increased risk of DME, whereas %lymphocytes is associated with a reduced risk of DME.

Herein, we combined a derivative of cisplatin (CP) and the chemosensitizer lonidamine (LND) to design an amphiphilic prodrug, in which the ratio of LND to cisplatin was fixed at 2:1. Diaminedichlorodihydroxyplatinum (DH-CP) is a hydrophilic cisplatin derivative. Due to its appropriate amphiphilicity, this prodrug could self-assemble into stable nanoparticles (denoted as LNP-NPs). Under the action of excessive glutathione (GSH) in tumor cells, DH-CP could be reduced to cytotoxic cisplatin. In addition, the released LND could inhibit the metabolic process of tumor cells, and improving the sensitivity of tumor cells to cisplatin. In vitro studies demonstrated that LNP-NPs displayed significantly cytotoxicity on breast cancer cells, and the cell viability after co-incubation for 48 h (CP 16 μg/mL) were 18.77% (MCF-7) and 20.01% (EMT6), respectively. LNP-NPs could also significantly inhibit the growth of MCF-7 tumor-like spheroids, which were realized through the high coordination and cooperation between CP and LND. Therefore, the carrier-free drug delivery system based on LND and DH-CP is expected to achieve a good synergistic anti-tumor effect.

The k nearest neighbor (KNN) classifier is one of the well-known instance-based classifiers. Nevertheless, the low efficiency in both running time and memory usage is a great challenge in the KNN classifier and its improvements due to noise and redundant samples. Although hybrid instance reduction approaches have been postulated as a good solution, they have the following issues: (a) adopted edition methods in existing hybrid instance reduction approaches are susceptible to harmful samples around the tested sample; (b) existing hybrid instance reduction approaches retain many internal samples, which contributes little to the classification accuracy and (or) leading to the low reduction rate; (c) existing hybrid instance reduction approaches rely on too many parameters. The chief contributions of this article are that (a) a novel heuristic hybrid instance reduction approach based on adaptive relative distance and k-means clustering (HIRRDKM) is proposed against the above issues; (b) a novel concept, i.e., the adaptive relative distance, is first proposed and calculated for each sample; (c) a novel edition method based on adaptive relative distance (EMARD) in HIRRDKM is second proposed to filter out harmful samples; (d) a novel condensing method based on adaptive relative distance and k-means clustering (CMRDKM) in HIRRDKM is third proposed to obtain condensed borderline samples from the training set without harmful samples. Experiments have proved that (a) HIRRDKM outperforms 6 state-of-the-art hybrid instance reduction methods on real data sets from various fields in weighing reduction rate and classification accuracy of KNN-based classifiers; (b) the running time of HIRRDKM is competitive.