面向动态交通流量预测的自适应图注意Transformer
|
|
刘宇轩,刘毅志,廖祝华,邹正标,汤璟昕
|
Adaptive graph attention Transformer for dynamic traffic flow prediction
|
|
Yuxuan LIU,Yizhi LIU,Zhuhua LIAO,Zhengbiao ZOU,Jingxin TANG
|
|
| 表 5 PEMS08数据集上不同变体的RMSE和MAE对比结果 |
| Tab.5 Comparison results of RMSE and MAE for different variants on PEMS08 dataset |
|
| 模型 | Tp=15 min | | Tp=30 min | | Tp=45 min | | Tp=60 min | | RMSE | MAE | | RMSE | MAE | | RMSE | MAE | | RMSE | MAE | | w/o MH | 22.96 | 16.78 | | 25.61 | 19.32 | | 29.25 | 20.45 | | 34.73 | 26.33 | | w/o LH | 24.08 | 17.58 | | 24.26 | 18.06 | | 27.13 | 19.08 | | 31.93 | 22.14 | | w/o LM | 29.72 | 20.47 | | 26.34 | 19.86 | | 28.71 | 19.43 | | 27.65 | 20.42 | | w/o adp | 23.73 | 16.97 | | 24.58 | 18.75 | | 27.16 | 19.14 | | 30.69 | 20.04 | | w/o dis | 27.38 | 18.12 | | 28.96 | 18.96 | | 30.12 | 20.43 | | 31.25 | 21.54 | | MSAGAFormer | 21.87 | 16.05 | | 23.87 | 17.26 | | 25.13 | 17.76 | | 26.63 | 19.14 |
|
|
|