Code Monkey home page Code Monkey logo

urfc2019_preliminary's Introduction

URFC2019_Preliminary

2019 The 5th Baidu & XJTU Big Data Contest The First IKCEST “The Belt and Road” International Big Data Contest

线上测试准确率

  • 单模型单次训练, online:0.7117
  • 单模型十折融合, online:0.7199 排名:19

网络结构及改进

  • 复现ABadCandy Pytorch版本的baseline,online:0.6678。
  • 采用 image,visit 双网络,分别生成256维特征,concat 后得到512维特征全连接到9维输出。
  • image 采用 imagenet 上预训练的 ResNet-50 提取特征。
  • visit 采用手动搭建的无预训练的简单网络提取特征(2个inception module和2个residual block)。
  • visit 先经过简单预处理保存为24×182×1的ndarray格式,我们将visit经过归一化,标准化等5种不同的预处理,得到shape为24×182×5的数据,送入visit网络。
  • 对image和visit均进行了数据增强,应对过拟合(visit增强的方式:变换到0-255的像素范围内将其当作图片进行数据增强)。
  • visit和image网络采用不同的学习率。
  • 可以选择使用mixup,默认未使用。

运行环境

  • Ubuntu16.04
  • Pytorch 1.1.0
  • NVIDIA GTX 1080Ti×2

运行说明

1. 准备数据

同ABadCandy Pytorch版本的baseline

cd URFC2019_Preliminary
mkdir data
cd data
mkdir train test  #创建训练image文件夹train, 测试图片文件夹test
mkdir -p npy/train npy/test  #创建训练visit文件夹npy/train, 测试visit文件夹npy/test

分别将去噪后的image和ndarray格式的visit数据放入相应文件夹中。

或者直接从百度网盘链接直接下载我们提供的已经预处理过的数据,解压到工程目录下

提取码:ha81

2. 运行程序

数据准备好后,运行multimain.py文件即可进行单模型单次训练。

  • data/ 数据文件夹
  • preliminary/ 存放初赛数据csv文件
  • multimain.py 主程序
  • multimodal.py 网络模型
  • preprocess.py 数据增强
  • config.py 参数配置
  • utils.py 通用模块

注:

  1. 可在config.py中更改参数配置。
  2. 可在multimodal.py中更改网络结构。
  3. 默认采用sgd训练60epoches,带warmup的cos学习率策略,训练时间大约3hours,为加速可以选择更换优化器,减少epoch数。

如果您觉得有用,还请多多star,感谢!

urfc2019_preliminary's People

Contributors

xjtuwxliang avatar

Stargazers

deng avatar

Watchers

James Cloos 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.