The Google "Crash Course on Machine Learning" and why YOU should do it

vorsprung profile image vorsprung ・2 min read

"While there have been advances in artificial intelligence (AI) this year, it's poised to skyrocket in 2019.." So a dozen articles a week begin. But what does this mean to me?

I basically understand that "General AI" or computers that can comprehend a wide range of topics and interests, just like people do, is still a long way off

However, Machine Learning is here right now. Teaching a model with real-world data and then predicting events using statistics is viable.

I don't pretend to understand the ins and outs of the complex mathematics that underpin machine learning. However, is there something in it for an average guy with a bit of experience in managing systems and software development?

I decided to do a course to see!

The course I picked was the Google "Crash Course in Machine Learning"

I'd already tried a few things with Tensorflow and this course used Tensorflow as it's a basic framework

The course covers the high-level theories that are used to build models, gather data and refine the predictions. There are about 20 hours of videos and practical exercises using a Jupiter-notebook alike Machine Learning playpen

To me the best realisation was in one of the later sections, summarised in this diagram

The tiny blue section is the super clever stuff with lots of complex maths

The rest of the sections are, basically, tradition data handling practices. If you have experience with moving data about, cleaning it and other usual validation processes then most of the mysterious world of Machine Learning is something you already know!

So it's well worth doing this course to see how your existing skills and experience fit in with the up and coming Machine Learning scene

Posted on Nov 9 '19 by:

vorsprung profile



Many disks offline again


markdown guide

I think Andrew Ng's course on coursera is a better learning resource. Google has enough influence without dictating the first crucial introduction to a subject.

Since you mentioned tensor flow: while it's a nice tool, it makes you skip a step or two, which is detrimental to your initial understanding of the subject.

But yeah, most of data science job consists of fixing shoddy data management in my experience...


agree, that intro from google is really nice. going over it then on to your fav course goes a long way into consolidating your understanding. they really put the beans in it!

edit: i wanted to put a hashtag before agree but it turned out like THAT


Wow, thanks for sharing this - never heard about Google's course, seems like the exact 101 course I was looking for


Thanks for article; I'll probably try it, although it is not my primary goal to focus.


up and coming Machine Learning scene

that has already proved that it needs to be stopped


Maybe you are talking about controversies with AWS Rekognition or Palantir? The technology is out of the pandora's box and we have to understand it and use it correctly. Or not use it in some contexts.

The Google course does have a whole section on selection bias and how and why to avoid it


I think, that what he means is basically:

People: Correlation is not causation
Machien Learning: ........ really?