Understanding Machine Learning: From Theory to Algorithms (Downloadable Book)

cs.huji.ac.il

80 points·namin··3 comments

When you guys are studying and learning from this, what is your strategy? Do you read the entire book online? Do you work through it and implement various things? ...?

Just like with learning how to program, it's ideal to have a personal project to go along with a book. This way when you dive in to the book(s) the knowledge learned will be more likely to stick.

This will leave holes in ones knowledge, because different projects have different challenges. Therefore, it's helpful to hold onto good books and come back to them when going over other projects, reviewing and learning even more useful information.

As for this book, I haven't read it, but it looks good and gentle, so there is potentially less room for holes when reading this book. On the other end, when doing a project, there is a higher chance of needing to dive into secondary material to learn edge cases. It's a common trade off. All in all, this book looks quite good from a ten thousand foot view.

I downloaded the PDF and looked at the beginning of two random chapters. My first impression is that the authors define machine learning formally, but fairly simply and then offer practical material. Not really fair to offer an opinion after a 15 minute look, but it looks like a good book.

Reply on news.ycombinator.com