Mathematics for machine learning and neural networks with a school knowledge base?

Question

go

Greetings!
I am a PHP developer and an IT student during my first semester abroad. I recently decided to start learning machine learning and neural networks as a hobby. I started watching ShAD courses (and some others) and realized that I don't understand anything about the mathematics that they give you there. Small ShAD I understand normally, but this is not enough.
The problem is that universities abroad do not provide a sufficient amount of knowledge in mathematics as in Russian universities, and I doubt that they will give something more than the Russian school mathematics base.
If any books/courses on math that would be useful for machine learning and especially for neural networks, for people who have completed grade 11 and have a good base in school mathematics (at the USE level). The language of the courses/books is Russian or English . Also, if you know any courses on machine learning and neural networks for "dummies" and which not only give a lot of theory, but also show it in practice (programming in python or any other language), let me know!
I already searched for answers on the toaster and in Google, but mostly books / courses for complete zeros, or for an advanced level.
Thank you!

Answer 1

The terminology is confusing. map is not a queue, channel is a queue. A queue can be made from slice or list by implementing the Put and Get methods. map is needed for direct access by name, it's cheaper and more deterministic to iterate in a slice loop.

Answer 2

Program:
https://download.cdn.yandex.net/shad/shad_program_...
In my case, it looks something like this:
PS: the main thing is to understand that knowledge is a process, if you lack some knowledge, then you can always catch up, but it takes time, and the less you know, the more time it takes. Two years ago I knew mathematics at the level of the 6th grade of school, now I study at the ShAD in the 2nd year and read the Deep Learning Book (although in some places I have to dig into textbooks).
PPS: when immersed in some area, the first book you read becomes the table of contents for this area of ​​knowledge and you will go further based on it, so books accumulate, they are not always read from cover to cover.

Answer 3

I recommend an introductory machine learning course from Andrew Ng on the courser: https://www.coursera.org/learn/machine-learning/ho...
The course assumes a not very deep knowledge of mathematics from the student: for mastering it is not necessary to know what function limit and derivative or some concepts from tv and ms. All algorithms are told on the fingers, the simplest intuition is given for understanding. The only thing is that you will have to program in matlab or octave in order to hand in assignments for assessment.
On the other hand, the same courser has a specialization from MIPT/Yandex, it provides all the basic knowledge of mathematics that will be needed during the course. The theory is relatively well supported by examples. The code must be written in Python. The main disadvantage of the course is that you will not implement any algorithms, the emphasis is on the use of ready-made ones in the sklearn library (unlike the same course by Andrew Yn).
Edited: I forgot to add that there is a good book on machine learning algorithms in general - www-bcf.usc.edu/~gareth/ISL/. Introduction to statistical learning. It seems to be considered a good choice for introduction to the area.

Answer 4

Almost all mathematics for neural networks and machine learning is TV / statistics, linear algebra and some analysis (up to partial derivatives approximately). I believe that mathprofi.ru will cover all the needs at the initial stage - understanding the courses. If you have a lot of time or prefer to learn from books, then on TV / ms you can take "Brodsky I with statistics, the probability of combinatorics", if on hardcore, then Feller's two-volume book. These books are very practical.
According to mathematical analysis, one can take Fichtenholtz, he wrote very clearly, but very voluminously.
On a linear basis, I don’t know what to advise, generally recognized Soviet books such as Kostrikin, Manin (recommended in the list of ShAD questions) are just brainwashing. Tyrtyshnikov Matrix analysis and linear algebra speak well. As an alternative - a course on the steppe dedicated to LA.

Answer 5

https://stepic.org/course/%D0%9D%D0%B5%D0%B9%D1%80...

Answer 6

Can I ask what country of study and university?