How to make your first neural network?

Question

Python

GOOD TIME EVERYONE!
Briefly about me :
I am 15 years old !
I know python3, understand php a little.
I can work with SQL :)
I can write a website html, css . - But this topic does not interest me. I studied it because it was necessary sometime :)
======================================== ===============================================
Interested in the topic machine learning .
But according to examples from the Internet, I can’t understand anything what is there and how it works and what results it gives out.
Is there something like courses on neural networks without water :)
Because I watched everything. The Internet is too complicated, YouTube has too much water.
================================================= =======================================
Interested in either personal experience or a site where everything is simply and briefly explained.

Answer 1

To master machine learning, you need a good base, foundation, possession of a computer.
You ask that a difficult topic be presented to you on a silver platter, chewed. It doesn't work like that.
You need to start from afar, you won’t be able to master it from a swoop.
PS in general, that's why there are no young Padawans in machine learning

Answer 2

Just buy this book and repeat exactly everything that is written in it, then you can already change something as you wish https://www.ozon.ru/context/detail/id/141796497/

Answer 3

Try it - https://www.kaggle.com/learn/deep-learning
There are a lot of problems and examples of their solutions on a variety of topics and levels on kaggle - study them and try to understand, repeat. The probable outcome - you will understand that there is not enough foundation. Personally, I started with a specialization from Yandex and MIPT on coursera. To enter the ML - fire.

Answer 4

Well done, you Nikita! at the age of 15, I already stir up my neural network (but to be honest, I even envy a little that I personally have not been 15 for a long time).
However, I approx. now at the same stage, therefore, from my point of view, IMHO,
in order to master MO
1. you need to program very well, i.e. solve problems and write good/working python scripts in order to understand something, how, i.e. in addition to the base, you also need to know algorithms / patterns / OOP
well 2. know mathematics well: matan, linal, statistics, probability theory, discretization, in order to understand what is happening and be able to model the date (by the way, to understand the simulation, you can even read books on excel to understand at least the concept of what modeling is)
3. further study all the auxiliary frameworks for ML, i.e. such as pandas, matplotlib, scikitlearn, to work with date and its suitability
4... and only then you will get to the ability to work with TensorFlow, Keras, PyTorch and other ML libraries and tools
try somewhere download or buy Jose Portilla lectures on Udemy, it has everything, but also everything is a bit too superficial, plus really start doing MO IMHO you still need to go through points 1.+2.+3 very clearly.
There are a lot of O'Reilly books on the root tracker, too,
good luck!

Answer 5

There is a good book Deep Learning in Python . The author presents the material quite accessible to the reader without any special mathematical background (high school level is quite enough). Try to read a book and do something, maybe it will captivate you.