(1) DIY applied machine learning involves 4 practical steps: learning theory concepts, coding basics in Jupyter notebooks, setting up a machine like a Raspberry Pi, and finally building AI projects. (2) The document recommends starting with free online resources to learn theory, then using Google Collab or Azure Notebooks for coding practice. (3) It suggests getting a Raspberry Pi to run code on locally and learn Linux basics. (4) Finally, one can use free cloud computing and build visual or audio AI projects on a Raspberry Pi or commercial products like Amazon Echo.
NO1 Top Black Magic Specialist In Lahore Black magic In Pakistan Kala Ilam Ex...
DIY Applied Machine Learning
1. DIY Applied Machine Learning:
Practical and Collaborative Method to Jump Start into Machine Learning with Jupyter
Notebooks and Google Collab
Tarek Hoteit - PhD, Director of IT @ Thomson Reuters
http://tarek.computer
IDEAS Conference Dallas 2018
2. About me - Tarek Hoteit
First encounter with a “computer” was
at 6 years of age with Atari Pacman
Still fascinated with 8-bit old-school programming &
retro games 34 years later
My first lines of code when I was 8 years old:
10 Print “hello”
20 goto 10
Got obsessed with text
adventure games which lead
to natural language
processing and terminal-
based coding
Once in 80s encountered an article on neural networks for
Commodore 64 that made no sense to me .. but that curiosity
stayed deep in my brain
... Continue to be passionate about computers: coding,
tinkering with IoT devices, playing retro games, running
servers...
3. Curiosity drove me to love computers and got me into artificial intelligence, machine learning, etc.
How about you? How do you get into AI by yourself?
https://image.slidesharecdn.com/100608pkn-100608043211-phpapp01/95/curiosity-
empathy-and-imagination-6-728.jpg?cb=1309980347
4. Traditionally you go .... on Twitter
- FOLLOW Machine Learning experts on Twitter
- READ tweets about new implementations
9. You ask yourself....
- How do I start on my own?
- What project can I do?
- How can I show my work?
- How do I setup the environment for myself?
- What is a great use case to build on?
- Which framework would I use?
- Which language would I choose?
- Will I get employed with my knowledge?
- Others know more than me?
- And on and on and on...
15. And the way to do it yourself with machine learning?
4 STEPS
16. DO IT YOURSELF with machine learning - step #1: theory
● Understand what is supervised, unsupervised, deep learning, and reinforcement learning concepts
○ Check mindmap summarizing concepts from data analysis to deep learning here
○ Business people can read this, coders can read that, and scientists can read these
arxiv.org
17. DO IT YOURSELF with machine learning - step #2: code
● Start with a blank but preset Jupyter notebook - use Google Collab https://colab.research.google.com or
Microsoft Azure Notebooks https://notebooks.azure.com/
● Learn some critical basics using Jupyter notebook as a playground - check https://jupyter-
notebook.readthedocs.io/en/stable/
● Basic Python learning - don’t go deep.. Just how to start writing basic lines of code - check
https://www.python.org/about/gettingstarted/
● Delimited files concept / CSV how to import and how to export - love Pandas! - check
https://pandas.pydata.org/pandas-docs/stable/io.html
● Package installation - PIP & setting up requirements.txt for storing all packages to install - check
https://packaging.python.org/tutorials/installing-packages/
● Understand and use Python virtual environments - check http://docs.python-
guide.org/en/latest/dev/virtualenvs
● Learn markdown syntax for content publishing https://github.com/adam-p/markdown-
here/wiki/Markdown-Cheatsheet
19. DO IT YOURSELF with machine learning - step #3: machine
● Skip cloud computing (AWS, Azure, gcloud) until Step 3.. Learn to run Linux in your
home!
● Pick up a Raspberry PI & SD Card from a local store.
● Install Linux (Raspbian distribution - check this guide ) and learn the basics of:
○ Terminal commands : ls, cd, md, ssh, etc.... check Linux basics on http://www.aboutdebian.com/linux.htm
● Setup your coding editor: just pick one... check Quora for ideas or just pick PyCharm
for Python or Java Eclipse or Sublime Text2 or VIM if you are brave and patient.
● Learn database basics such as Postgres or MYSQL.
● Get yourself comfortable with your programming language of choice. I personally love
Python (tutorial) but you may want Java (tutorial) or C# (quickstart) or whatever.
● Practice cloning existing code repositories from GitHub using GIT (tutorial)
● Algorithms can be found on github.com or gitxiv.com
Raspberry PI waiting for you
at local Microcenter store or
on Amazon
20. Host your IPython notebook on your Raspberry PI
Some how to links:
https://blog.domski.pl/ipython-notebook-server-on-raspberry-pi/ or
https://www.raspberrypi.org/forums/viewtopic.php?t=130450&p=870964
21. DO IT YOURSELF with machine learning - step #4: AI finally!
- Learn to setup free-tier Amazon AWS instance (1 year free tier) or Microsoft
Azure (free trial) or Google Cloud (free tier) for cloud computing setup and
machine learning deployments
- For Visual DIY AI projects:
- Buy yourself a Raspberry Pi Camera (https://www.raspberrypi.org/products/camera-module-v2/)
- Check Google AIY Projects https://aiyprojects.withgoogle.com/vision/
- Check Raspberry PI projects ideas here or here
- For Audio DIY AI projects:
- Get yourself an Amazon Echo from Amazon or any store, learn to build Alexa Skills (getting
started)
- Or try Google AIY project with Google Now integration - https://aiyprojects.withgoogle.com/voice/
22. Build various kits
using Google AIY
or Raspberry PI
Or run reinforcement learning projects using games!
Check https://keon.io/deep-q-learning/ and
https://github.com/DanielSlater/PythonDeepLearningS
amples
23. Now pick a project visual or audio for your
home... and for your family....
25. To connect with Tarek Hoteit
Blog: https://tarek.computer (includes copy of the slides)
Twitter: @hoteit
LinkedIn: https://www.linkedin.com/in/hoteit/
Email: tarek.hoteit@tr.com