Sun Yat-sen University

About: Sun Yat-sen University is a education organization based out in Guangzhou, Guangdong, China. It is known for research contribution in the topics: Population & Cancer. The organization has 115149 authors who have published 113763 publications receiving 2286465 citations. The organization is also known as: Zhongshan University & SYSU.

Self-supervised Graph Learning for Recommendation

Abstract: Representation learning on user-item graph for recommendation has evolved from using single ID or interaction history to exploiting higher-order neighbors. This leads to the success of graph convolution networks (GCNs) for recommendation such as PinSage and LightGCN. Despite effectiveness, we argue that they suffer from two limitations: (1) high-degree nodes exert larger impact on the representation learning, deteriorating the recommendations of low-degree (long-tail) items; and (2) representations are vulnerable to noisy interactions, as the neighborhood aggregation scheme further enlarges the impact of observed edges. In this work, we explore self-supervised learning on user-item graph, so as to improve the accuracy and robustness of GCNs for recommendation. The idea is to supplement the classical supervised task of recommendation with an auxiliary self-supervised task, which reinforces node representation learning via self-discrimination. Specifically, we generate multiple views of a node, maximizing the agreement between different views of the same node compared to that of other nodes. We devise three operators to generate the views -- node dropout, edge dropout, and random walk -- that change the graph structure in different manners. We term this new learning paradigm as \textit{Self-supervised Graph Learning} (SGL), implementing it on the state-of-the-art model LightGCN. Through theoretical analyses, we find that SGL has the ability of automatically mining hard negatives. Empirical studies on three benchmark datasets demonstrate the effectiveness of SGL, which improves the recommendation accuracy, especially on long-tail items, and the robustness against interaction noises. Our implementations are available at \url{this https URL}.

The role of the microbiome in NAFLD and NASH

Abstract: Nonalcoholic fatty liver disease (NAFLD) is the hepatic manifestation of cardiometabolic syndrome, which often also includes obesity, diabetes, and dyslipidemia. It is rapidly becoming the most prevalent liver disease worldwide. A sizable minority of NAFLD patients develop nonalcoholic steatohepatitis (NASH), which is characterized by inflammatory changes that can lead to progressive liver damage, cirrhosis, and hepatocellular carcinoma. Recent studies have shown that in addition to genetic predisposition and diet, the gut microbiota affects hepatic carbohydrate and lipid metabolism as well as influences the balance between pro-inflammatory and anti-inflammatory effectors in the liver, thereby impacting NAFLD and its progression to NASH In this review, we will explore the impact of gut microbiota and microbiota-derived compounds on the development and progression of NAFLD and NASH, and the unexplored factors related to potential microbiome contributions to this common liver disease.

The State of the World's Mangrove Forests: Past, Present, and Future

Abstract: Intertidal mangrove forests are a dynamic ecosystem experiencing rapid changes in extent and habitat quality over geological history, today and into the future. Climate and sea level have drastical...

Comparison of Ultrasound Elastography, Mammography, and Sonography in the Diagnosis of Solid Breast Lesions

Abstract: Objective The purpose of this study was to evaluate the value of ultrasound elastography (UE) in differentiating benign versus malignant lesions in the breast and compare it with conventional sonography and mammography. Methods From September 2004 to May 2005, 296 solid lesions from 232 consecutive patients were diagnosed as benign or malignant by mammography and sonography and further analyzed with UE. The diagnostic results were compared with histopathologic findings. The sensitivity, specificity, accuracy, positive and negative predictive values, and false-positive and -negative rates were calculated for each modality and the combination of UE and sonography. Results Of 296 lesions, 87 were histologically malignant, and 209 were benign. Ultrasound elastography was the most specific (95.7%) and had the lowest false-positive rate (4.3%) of the 3 modalities. The accuracy (88.2%) and positive predictive value (87.1%) of UE were higher than those of sonography (72.6% and 52.5%, respectively). The sensitivity values, negative predictive values, and false negative rates of the 3 modalities had no differences. A combination of UE and sonography had the best sensitivity (89.7%) and accuracy (93.9%) and the lowest false-negative rate (9.2%). The specificity (95.7%) and positive predictive value (89.7%) of the combination were better, and the false-positive rate (4.3%) of the combination was lower than those of mammography and sonography. Conclusions In a clinical trial with Chinese women, UE was superior to sonography and equal or superior to mammography in differentiating benign and malignant lesions in the breast. A combination of UE and sonography had the best results in detecting cancer and potentially could reduce unnecessary biopsy. Ultrasound elastography is a promising technique for evaluating breast lesions.