Skip to content

jasonz5/STM-MAE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Spatio-Temporal-Agnostic Masked Pre-training

Dependencies

The code is implemented based on STD-MAE, which is realized on BasicTS and STEP.

Thinking 1: ST-Agnostic Masking

  • STEP和STD-MAE中采用了MAE,预测效果好于之前的SOTA方法,俩者分别采用是时间掩码和时空解耦掩码。
  • 而Kaiming的Video-MAE工作表明视频数据ST-Agnostic掩码效果更好。
  • Traffic和Video本质都是时空数据,这一部分尝试验证ST-Agnostic掩码的效果。

STM-MAE methodology. Video-MAE work. Challenges of STM-MAE.

Thinking 2: LargeST Transfer

  • 相比于PEMS0x数据集,LargeST中提出了在时空维度都很大的数据集,包含了加州5年的Traffic Datasets
  • 该部分要想要验证下大数据集相比于PEMS0x在城市迁移上的提升效果。

LargeST datasets. PEMS04 Transfer. PEMS04 Simple Transfer. Transfer Analysis.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published