GitHub is where people build software. More than 150 million people use GitHub to discover, fork, and contribute to over 420 million projects.
References STGCN in traffic:https://github.com/FelixOpolka/STGCN-PyTorch. Seq2Seq in Neural machine translation:https://www.tensorflow.org/tutorials/text/nmt_with_attention. Releases No releases published
python train.py --save_dir <save-dir> --demo_file /mimic_admission_demo.csv --edge_modality 'demo' --feature_type 'multimodal' --ehr_feature_file <ehr-feature-dir>/ehr_preprocessed_seq_by_day_cat_embedding.pkl \ --edge_ehr_file <ehr-feature-dir>/ehr_preprocessed_seq_by_day...
Code for our VLDB'22 paper Decoupled Dynamic Spatial-Temporal Graph Neural Network for Traffic Forecasting. - GitHub - GestaltCogTeam/D2STGNN: Code for our VLDB'22 paper Decoupled Dynamic Spatial-Temporal Graph Neural Network for Traffic Forecasting.
The code is available at https://github.com/Frank-Wang-oss/FCSTGNN. Dataset We use three datasets to evaluate our method, including C-MAPSS, UCI-HAR, and ISRUC-S3. C-MAPSS You can access here, and put the downloaded dataset into directory 'CMAPSSData'. For running the experiments on C...
github 上边暂时未找到开源 本文创新点: 提出了一种新的具有位置注意机制的图神经网络层,以更好地聚合邻近道路的交通流信息; 结合了RNN和Transformer,以捕获局部和全局的时间依赖性; 提出了一个新的时空GNN框架STGNN,该框架专门用于建模具有复杂拓扑和时间依赖性的系列数据 问题描述 每一个交通流网络可以被定义为G ...
NYTV project(based on spark and D2STGNN. Contribute to oneJue/NYTV development by creating an account on GitHub.
•代码:https://github.com/Frank-Wang-oss/FCSTGNN 给大家介绍一篇由新加坡科技研究局(A*STAR)与新加坡南洋理工大学联合发表的多元时间序列领域文献,目前已被AAAI2024录用。 01 研究背景 多元时间序列数据具有序列性和多源性,这使得数据内在地表现出空间-时间(ST)依赖性。这些依赖性包括时间戳之间的时间相关性以及...
原文: https://dl.acm.org/doi/abs/10.14778/3551793.3551827 代码:https://github.com/zezhishao/D2STGNN1 本文创新点1.1 现有交通预测模型存在的问题最近提出的一些用来解决交通预测问题的时空图神经网络模型…
北京出租车数据集,郑宇,"BJ15_M32x32_T30_InOut.h5",原始数据shape=(5596,2,32,32),"2"代表出In/Out两种流量。"32,32"代表网格化地图形状。 备注:数据应用在ST-ResNet(AAAI17,郑宇的经典,该领域的里程碑)中。 NYC-Taxi 纽约出租车数据集,"volume.train.npz",原始数据shape=(1920,10,20,2),"10,20...