MahatmaSun1 / SaConvSLTM

Self-Attention ConvLSTM for Spatiotemporal Prediction, described in `https://ojs.aaai.org//index.php/AAAI/article/view/6819`, test on MovingMNIST.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

overview

  • Self-Attention ConvLSTM for Spatiotemporal Prediction
  • SaConvSTLM described in this paper
  • test on MovingMNIST dataset

project structure

  • load_data.py

    包含用于下载数据集、对数据集进行预处理和保存预测图片到文件夹的实用函数

  • ConvSTLM_main.py

    使用 ·tensorflow.keras.layers.ConvLSTM2D 搭建的网络模型及其训练测试

  • SaConvSTLM.py

    基于tensorflow实现了cell SaConvLSTM2DCell和 layer SaConvLSTM2D,layerSaConvLSTM2D可以直接作为模型中的一个 layer 使用

  • SaConvSTLM_main.py

    使用 自定义的SaConvLSTM层搭建的网络模型及其在MovingMNIST 上的训练测试

    模型结构

    SaConvLSTM_model_structure

SaConvLSTM structure

  • described in this paper

  • overall pattern

    image-20210317231218553

  • self_attention memory module

    image-20210317231301247

About

Self-Attention ConvLSTM for Spatiotemporal Prediction, described in `https://ojs.aaai.org//index.php/AAAI/article/view/6819`, test on MovingMNIST.


Languages

Language:Python 100.0%