How Much Math Do You Need to Know for Machine Learning?
If you know your single-variable calculus, then it's about 70 hours on Math Academy.
Want to get notified about new posts? Join the mailing list and follow on X/Twitter.
There’s a lot of debate regarding how much math you need to know for machine learning.
Lots of the “Math for Machine Learning” textbooks out there are meant for academic researchers in machine learning theory, but are being passed off to students and engineers as the math that they need to know before studying machine learning.
For instance, the textbook Algebra, Topology, Differential Calculus, and Optimization Theory For Computer Science and Machine Learning weighs in at nearly 2,200 pages!
There are some humorous reactions in a Twitter thread referencing the book:
- "Where in the hell you're supposed to use Zorn's lemma in ML?" link
- "Great but it is missing probability and has a lot of topics rarely if ever needed in ML" link
- "Why do we need the following:
a) Spectral theorem
b) Spectral Graph Drawing
c) Cartan-Dieudonn´e Theorem
Where do we apply this?" link - "So, in my estimate, intersection over union < 0.1." link
(Translation: the amount of relevant information in the book is less than 10% of the total information in the book plus any relevant information that is not in the book.)
If you want to study machine learning, but don’t have the proper mathematical foundations, and don’t want to spend years working through a giant textbook filled with stuff that won’t actually come up in an interview or on the job, then check out Math Academy’s Mathematics for Machine Learning course.
The course can be completed in about 70 hours if you know your math up through single-variable calculus. (If not, then the diagnostic assessment will automatically figure out the lower math that you need to learn and add that to your learning plan.)
Want to get notified about new posts? Join the mailing list and follow on X/Twitter.