Search

Your search keyword '"Li, Yongyong"' showing total 666 results
Start Over Author "Li, Yongyong"
666 results on '"Li, Yongyong"'

18. GraphTheta: A Distributed Graph Neural Network Learning System With Flexible Training Strategy

Author

Liu, Yongchao, Li, Houyi, Zhang, Guowei, Zeng, Xintan, Li, Yongyong, Huang, Bin, Zhang, Peng, Li, Zhao, Zhu, Xiaowei, He, Changhua, and Chen, Wenguang

Subjects
Computer Science - Machine Learning, Computer Science - Distributed, Parallel, and Cluster Computing
Abstract

Graph neural networks (GNNs) have been demonstrated as a powerful tool for analyzing non-Euclidean graph data. However, the lack of efficient distributed graph learning systems severely hinders applications of GNNs, especially when graphs are big and GNNs are relatively deep. Herein, we present GraphTheta, the first distributed and scalable graph learning system built upon vertex-centric distributed graph processing with neural network operators implemented as user-defined functions. This system supports multiple training strategies and enables efficient and scalable big-graph learning on distributed (virtual) machines with low memory. To facilitate graph convolutions, GraphTheta puts forward a new graph learning abstraction named NN-TGAR to bridge the gap between graph processing and graph deep learning. A distributed graph engine is proposed to conduct the stochastic gradient descent optimization with a hybrid-parallel execution, and a new cluster-batched training strategy is supported. We evaluate GraphTheta using several datasets with network sizes ranging from small-, modest- to large-scale. Experimental results show that GraphTheta can scale well to 1,024 workers for training an in-house developed GNN on an industry-scale Alipay dataset of 1.4 billion nodes and 4.1 billion attributed edges, with a cluster of CPU virtual machines (dockers) of small memory each (5$\sim$12GB). Moreover, GraphTheta can outperform DistDGL by up to $2.02\times$, with better scalability, and GraphLearn by up to $30.56\times$. As for model accuracy, GraphTheta is capable of learning as good GNNs as existing frameworks. To the best of our knowledge, this work presents the largest edge-attributed GNN learning task in the literature., Comment: 19 pages, 13 figures, 8 tables

Published
2021

30. A novel ZRS variant causes preaxial polydactyly type I by increased sonic hedgehog expression in the developing limb bud

Author

Xu, Caixia, Yang, Xiaoming, Zhou, Hang, Li, Yongyong, Xing, Chao, Zhou, Taifeng, Zhong, Dongmei, Lian, Chengjie, Yan, Mei, Chen, Tao, Liao, Zhiheng, Gao, Bo, Su, Deying, Wang, Tingting, Sharma, Swarkar, Mohan, Chandra, Ahituv, Nadav, Malik, Sajid, Li, Quan-Zhen, and Su, Peiqiang

Subjects
Human Genome, Genetics, 2.1 Biological and endogenous factors, Aetiology, Animals, Caco-2 Cells, Disease Models, Animal, Female, Gene Knock-In Techniques, HEK293 Cells, Hedgehog Proteins, Heterogeneous-Nuclear Ribonucleoprotein K, Humans, Introns, Limb Buds, Male, Membrane Proteins, Mice, Mice, Transgenic, Pedigree, Polydactyly, Polymorphism, Single Nucleotide, Thumb, Up-Regulation, preaxial polydactyly, ZRS, sonic hedgehog, gene regulation, Clinical Sciences, Genetics & Heredity
Abstract

PurposePreaxial polydactyly (PPD) is a common congenital hand malformation classified into four subtypes (PPD I-IV). Variants in the zone of polarizing activity regulatory sequence (ZRS) within intron 5 of the LMBR1 gene are linked to most PPD types. However, the genes responsible for PPD I and the underlying mechanisms are unknown.MethodsA rare large four-generation family with isolated PPD I was subjected to genome-wide genotyping and sequence analysis. In vitro and in vivo functional studies were performed in Caco-2 cells, 293T cells, and a knockin transgenic mouse model.ResultsA novel g.101779T>A (reference sequence: NG_009240.2; position 446 of the ZRS) variant segregates with all PPD I-affected individuals. The knockin mouse with this ZRS variant exhibited PPD I phenotype accompanying ectopic and excess expression of Shh. We confirmed that HnRNP K can bind the ZRS and SHH promoters. The ZRS mutant enhanced the binding affinity for HnRNP K and upregulated SHH expression.ConclusionOur results identify the first PPD I disease-causing variant. The variant leading to PPD I may be associated with enhancing SHH expression mediated by HnRNP K. This study adds to the ZRS-associated syndromes classification system for PPD and clarifies the underlying molecular mechanisms.

Published
2020

31. Prediction of height of water flowing fractured zone based on improved SSA to optimize BP neural network

Author

WANG Yaoguo, LI Yongyong, and GUO Tao

Subjects
water flowing fractured zone, bp neural network, sparrow search algorithm(ssa), tent chaotic mapping, gaussian mutation, Mining engineering. Metallurgy, TN1-997
Abstract

Aiming at the problem that it is difficult to accurately predict the height of water flowing fractured zone in coal mining subsidence disaster early warning modeling, a prediction model based on improved sparrow search algorithm (SSA) to optimize BP neural network is proposed to predict the height of water flowing fractured zone. Add Tent chaotic mapping to the standard SSA to initialize the population, improve the uniformity and diversity of population distribution, and enhance the global search ability of SSA; with the help of Gaussian mutation and Gaussian perturbation algorithm, the ability of SSA to escape from local optimization is improved; dynamic step size is introduced to instead of random step size to improve the solution accuracy of SSA. Through practical application, the improved SSA-BP was compared with SSA-BP, PSO-BP, BP and the prediction results of GA-SVR model studied by predecessors. The results show that the average absolute error (MAE), average absolute percentage error (MAPE), root mean square error (RMSE) and determination coefficient(R2) of the improved SSA-BP prediction model are 1.23 m, 2.64%, 1.51 m and 0.985 respectively. It is better than other models and improves the accuracy of predicting the height of water flowing fractured zone.

Published
2023
Full Text
View/download PDF

42. Erythrocyte‐Derived Bioactive Nanovesicles Reverse the Immunosuppressive Function of Adenosine for Tumor Photoimmunotherapy

Author

Sun, Jiuyuan, primary, Liu, Yiqiong, additional, Zhao, Yuge, additional, Yin, Weimin, additional, Yang, Zichen, additional, Chang, Jiao, additional, Huang, Li, additional, Chen, Shiyu, additional, Zhi, Hui, additional, Xue, Liangyi, additional, Zhang, Xiaoyou, additional, Dong, Haiqing, additional, and Li, Yongyong, additional

Published
2024
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results