Deeplearning Get Started

A concrete get started for deep learning in Vthree.AI (Chinese required)

Prerequisite

Python2.7: http://www.runoob.com/python/python-tutorial.html

Python3.6: http://www.runoob.com/python3/python3-tutorial.html

Git: http://www.runoob.com/git/git-tutorial.html

.gitignore: https://git-scm.com/docs/gitignore

unix: http://linuxcommand.org/lc3_learning_the_shell.php

Documentation

Official Doc: https://mxnet.apache.org

Best Practice: https://mxnet.incubator.apache.org/faq/security.html

CheatSheet: https://github.com/anchen1011/Awesome-MXNet/blob/master/apache-mxnet-cheat.pdf

Very Good Tutorials

Web Tutorial: https://zh.gluon.ai/index.html

Video Tutorial: https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ

1. Install MXNet

http://mxnet.incubator.apache.org/install/

2. Get Started with an Example

http://mxnet.incubator.apache.org/test/tutorials/python/mnist.html

3. Hands-on

https://github.com/opringle/multivariate_time_series_forecasting

4. Math

Basics: https://zh.gluon.ai/chapter_appendix/math.html

Applied: https://github.com/zackchase/mxnet-the-straight-dope/blob/7a00d1ec253129d844055870d59266a8a502f5c4/chapter06_optimization/gd-sgd.ipynb

5. Actually Learn to Build Something

https://blog.csdn.net/u012436149/article/details/78047278

6.1 Focused Topics

NLP-toolbox: http://gluon-nlp.mxnet.io

NLP-tutorial: https://zh.gluon.ai/chapter_natural-language-processing/index.html

CV-toolbox: https://gluon-cv.mxnet.io

NLP-toolbox: https://zh.gluon.ai/chapter_computer-vision/index.html

6.2 Vthree.AI Specific Topics

CNN: https://zh.gluon.ai/chapter_computer-vision/bounding-box.html

LSTM: https://zh.gluon.ai/chapter_recurrent-neural-networks/lstm.html

7. Deliverables Guidelines

Training Procedure (optional)

Clear execution flow (example: https://github.com/opringle/multivariate_time_series_forecasting)

Inference Procedure

Clear handler, where input format is clearly defined. Should be runnable with:

>>> from inference_module import handler
>>> handler(input, context=None)

Dependency

Document all dependencies with version information (example: mxnet==0.12.3)

Environment

Document the environment information including machine image/AMI information