Giter Site home page Giter Site logo

pm_pred's Introduction

课设项目:

用 ConvLSTM 实现 $PM_{10}$$PM_{2.5}$ 的浓度分布预测

公式:

ConvLSTM:

与 ConvLSTM 论文中的实现方式不同。

$$ I_{t} = Sigmoid(Conv(X_t, W_{xi}) + Conv(H_{t-1}, W_{hi}) + b_i) $$

$$ F_{t} = Sigmoid(Conv(X_t, W_{xf}) + Conv(H_{t-1}, W_{hf}) + b_f) $$

$$ O_{t} = Sigmoid(Conv(X_t, W_{xo}) + Conv(H_{t-1}, W_{ho}) + b_o) $$

$$ \tilde{C_t} = Tanh(Conv(X_t, W_{xc}) + Conv(H_{t-1}, W_{hc}) + b_h) $$

$$ C_t = F_t \odot C_{t-1} + I_t \odot \tilde{C_t} $$

$$ H_t = O_t \odot Tanh(C_t) $$

$$ Pred = ReLU(Conv(X_t, W_{xx}) + Conv(H_t, W_{xh}) + b_{p}) $$

ConvGRU:

$$ R_{t} = Sigmoid(Conv(X_t, W_{xr}) + Conv(H_{t-1}, W_{hr})+b_r) $$

$$ Z_{t} = Sigmoid(Conv(X_t, W_{xz}) + Conv(H_{t-1}, W_{hz})+b_z) $$

$$ \tilde{H_t} = Tanh(Conv(X_t, W_{xh})+Conv(R_t \odot H_{t-1}, W_{rh}) + b_{h}) $$

$$ H_t = Z_t \odot H_{t-1} + (1 - Z_t) \odot \tilde{H_t} $$

$$ Pred = ReLU(Conv(X_t, W_{xx}) + Conv(H_t, W_{xh}) + b_{p}) $$

进度:

update 4/10:

实现了 ConvLSTM 和 ConvGRU 的基本框架,以及训练所需的代码

update 4/11:
  1. 实现了 use_random_iter = False 的训练方式
  2. 实现了运行自动保存日志的功能
  3. 实现了保存和加载功能
  4. 优化了随机采样功能的逻辑
  5. 完善了注释
update 4/28:
  1. 预实验结束,正式开始实验
update 5/4:
  1. 实验完成。
update 5/7:
  1. 效果图绘制完成。

任务:

关键任务:
  • 学习 WRF
  • 生成原始数据
  • 训练模型
  • 撰写论文
闲的没事:
  • 实现 Transformer 版本的网络

参考:

[ 1 ]: NIPS 2015: Convolutional LSTM Network

[ 2 ]: 动手学深度学习

[ 3 ]: ConvLSTM实现参考

pm_pred's People

Contributors

blathers23 avatar

Stargazers

 avatar mhuia avatar  avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.