Author: Cassie Kozyrkov (@kozyrkov, Twitter: @quaesita)

These walkthrough slides help guide you through the code demo I wrote with Brian Foo (@bkungfoo):

github.com/google-aai/sc17

They're a mix of:

• Screenshots taken while I walk through the Step-by-Step Deep Learning Tutorial.
• Bonus slides from my old lectures with ML hints, tips, summaries, and pitfall alerts.

My original day-long ML workshop ("Making Friends with ML") did not have a hands-on component, so the next step was to write code following the same step-by-step structure as the workshop. I had a lot of fun collaborating with the amazing Brian Foo (@bkungfoo) to make that code for you here. Then I captured screenshots to help you along in case you get stuck following our instructions. Hope you have fun learning some deep learning!

Open Demo00.pdf from this repo in one browser tab and the README.md from the Step-by-Step Deep Learning Tutorial in another tab. Whenever you see instructions in the README, follow along in the slides. If you get stuck, hopefully the slides will help you get unstuck. When you're done with Demo00.pdf, move to Demo01.pdf, and so on...

Open Demo00.pdf from this repo. Think of it as watching over my shoulder while I open my console, click around, and write code. When you're done with Demo00.pdf, move to Demo01.pdf, and so on... Enjoy!

I post info about talks and fun things I make for you on Twitter: twitter.com/quaesita

• ML course on YouTube
• ML course on Coursera
• Intro ML textbook (clear math, R code): ISL
• Intro ML guide (sense of humor, Python code): guidetodatamining.com/
• Intro ML book (older classic with lots of wisdom, uses WEKA examples): Data Mining by Ian Witten
• Online intro to neural networks: neuralnetworksanddeeplearning.com/
• Read more about neural networks here
• Interactive tinkering with neural networks at playground.tensorflow.org/
• Bayesian ML intro with Python code: Bayesian Methods for Hackers
• Reference book PDFs (can give beginners indigestion): Bishop, Elements of SL

These screenshots were captured in winter 2017/8, so there's no guarantee things still look the same whenever you find this, dear reader. We strive to continually improve, so future experiences on Google Cloud Platform might differ from my slides.

Advice and process is based on my experience as a data scientist. Naturally, different practitioners have their own preferred ways of doing things.