Does a specialist need math ml?

Question

Python

Hello . I was very interested in studying ML, AI, neural networks and everything connected with it.
Is it worth studying the mathematical disciplines necessary in this area (Linear algebra, probability theory, mathematical analysis)?
Or focus on learning Python and all the required libraries?
What is more profit for building a normal AI?
PS It is assumed that there are about 6-7 free hours per day.

Answer 1

I'm already tired of reposting this picture millions of times:

The main question is - do you want to get a job as an ML scientist or do you want to become a good specialist ?
Mathematics and programming languages ​​are just tools. The more competent a specialist is, the more tools he has in his arsenal.

Answer 2

Answer 3

For ML, first of all, you need to learn mathematics.
Mathematics for machine learning and neural networks with a school knowledge base?

Answer 4

A good ml-specialist needs to know perfectly well the school (more precisely, physics and mathematics lyceum) program in mathematics and computer science. Well, the program of the first two courses of the standard mathematics department is very desirable, at least for a four. For ML, specific mathematics is needed: set theory, mat. logic, number systems, fundamentals of number theory, combinatorics, graphs, theory. probabilities, mat. statistics, matan (in R^n), linear algebra, analytic geometry, calculus of variations. It is enough to know specific things.

Answer 5

No longer needed, ml-specialists randomly mix model-based solutions built in TensorFlow, trying to find the optimal one.

Answer 6

What is more profit for building a normal AI?

I draw attention to the word: NORMAL !
In understanding his work: so that without a PC, and only with a pencil and paper, you can create the desired AI.
And, yes: it is very simple if you have an understanding of what you are doing!
And no libs and pythons are needed!
They only add complexity to the AI ​​development process (but make it easier to transfer READY! logic from paper to PC, and again: only if there is an understanding!).

Answer 7

We open WIKIPEDIA. We read: To build such methods, the means of mathematical statistics, numerical methods, optimization methods, probability theory, graph theory, various techniques for working with data in digital form are used.
How, without mathematics, to know that's all that is described in the means - I honestly do not understand. At least there is even the word MATHEMATICAL.
Even fucking INS is a MATHEMATICAL model.