Machine learning - where to start as a programmer?

Question

Neural networks

Friends, I have recently become interested in machine learning. Mathematical preparation of the 9th grade of the school (the one that I more or less remember). He began to impudently storm books, courses and videos on relevant topics, but ran into a lack of mats. preparation. Understood, dug up a bunch of resources and lessons for vyshmat. And now I'm sitting, I don't know which side to approach.
Knowledgeable people, tell me, please, which sections in the vyshmat are needed at first for take-off.
Thank you.

Answer 1

At first, you need the following:
On linear algebra and derivatives, I can advise "All Higher Mathematics Volume I - Krasnov M., Kiselev A., Makarenko G., Shakin E., Zalyapin V". But there is a lot of extra for a beginner.
On statistics and probability theory, I can advise "an elementary course on probability theory and mathematical statistics - A. Borodin" up to page 100 will be enough.
My advice is not to dig into tutorials in the beginning. You can dig up good articles on these topics on the hub to read there. Ideally, it is best to study theory and practice in parallel.
Some ML books cover all of these topics. I can recommend Python Machine Learning (Sebastian Raschka). And if there are problems with the English language - Building machine learning systems in Python - Luis Pedro Coelho, Willy Richart.

Answer 2

Here's a pretty good source of information (especially if you're planning to work with Python).
For specific math, see the Math section near the end .

Answer 3

List of books, videos and courses on machine learning and mathematics, all in Russian. Large, high quality collection. Almost all pdf'ki of books are googled.
https://ru.stackoverflow.com/a/683632/1084

Answer 4

I don’t recommend a courser, in particular Andrew Ng, because at first everything is more or less, but the further, the less he explains and it becomes incomprehensible, not so much what and how, but why.
I like Vorontsov's lectures https://www.youtube.com/watch?v=qLBkB4sMztk
where you can also download the texts of the lectures, they complement each other perfectly, everything becomes more or less clear.

Answer 5

Start with the coursers. No matter how trite. An excellent course from Andrew Yn - https://www.coursera.org/learn/machine-learning
A good course from HSE and Yandex, in my opinion, it uses standard Python libs https://www.coursera.org/learn/ vvedenie-mashinnoe-...
There is a cycle of courses from the same Yandex and MIPT, but I did not go through it completely, I just started. It seems to me somehow less balanced than voiced.
https://www.coursera.org/specializations/machine-l...
And then practice. I stopped being interested in machine learning because there were no real projects with its application. I'm not interested in solving problems on Kaggle and similar resources.

Answer 6

The article is not new, but it may well help you navigate:
https://habrahabr.ru/company/mlclass/blog/252743/
In any case, you can write to the author and get more information on this issue. Good luck"

Answer 7

It will be useful to follow the news and new developments in the field of machine learning, AI and neural networks. In the group https://vk.com/neural_nets and the channel https://t.me/aichannel in telegram every day they publish news and articles, collections of textbooks. Try looking there.