A Review of Hands-On Machine Learning with Scikit-Learn, Keras & Tensorflow by Aurélien Géron
Major Topics Covered
What Is Machine Learning and why we use it
Hint: It is “Giving Computers the ability to Learn without being explicitly programmed” — Arthur Samuel 1959
An overview of the different types of Machine Learning
Like Supervised and Unsupervised Learning
A GREAT explanation on why data is so important to Machine Learning
You could use the most advanced methods of the day but unless you have good data your models won’t be great
Testing and Validation
Along with how we can avoid overfitting/underfitting
Notable Takeaways
The above picture gives a quick look at the sheer number of technologies present in any small facet of the tech industry, but this chapter does a good job of focusing on the fundamentals rather than the technologies and flashiness concepts in the industry. It breaks down Machine Learning into some of its most basic concepts that will generalize well to any future projects. Some of the most important being the difference between supervised & unsupervised learning and underfitting& overfitting along with the traits of good data and various methods used to test and validate the efficacy of ML Models.
Additionally, the author splits his time speaking about the different types of ML systems and talking about Data. I thought that this was excellent because it helps remind the audience how important data is to machine learning in a world full of shiny tools and algorithms.
My Thoughts
I personally thought this chapter was an excellent refresher to anyone like myself who has some, or even those with a lot of, experience in the world of Machine Learning, but I wouldn’t recommend it for an absolute beginner as a LOT of information is covered. If you find yourself in this camp I would recommend that you take a course that moves more slowly and in-depth through the fundamental concepts of Machine Learning.
Up until this point I had only been familiar with Batch Learning so I was excited to learn about Online Learning. Having come from a more individual assignment based background I have a large gap in my knowledge when it comes to different types of productionized ml projects and this was a nice taste in how different types of projects work.
Lastly, I learned about the No Free Lunch Theorem. This was developed by David Wolpert in 1996 and stated “that if you make absolutely no assumption about the data, then there is no reason to prefer one model over any other.” More simply this means that there is no silver bullet in Machine Learning. Some cases will require Deep Learning while others can be solved with a Random Forest or just a linear regression model can be the best for others. So, when developing ML projects you need to 1. know your data and 2. attempt to pick the right model for the data at hand rather than just the newest or fanciest model available.
Thanks for reading!
If you have any questions or feedback please reach out to me on twitter @wtothdev.
Additionally, I wanted to give a huge thanks to Aurélien Géron for writing such an excellent book. You can purchase said book here (non-affiliate).
Disclaimer: I don’t make any money from any of the services referenced and chose to read and review this book under my own free will.
