Genetic algorithm. How is life done?

Question

Software testing

Hello. Today I asked questions and received quick answers, a very pleasant forum.
I'll tell you a little about myself.
I've been programming for about 5 years 3 of them in C # I always liked making bots, but lately I've been looking a lot about machine learning and came across a genetic algorithm.
I figured out and implemented a genetic algorithm that collects the specified word, but it looks more like a selection with an estimate, but not the point.
I came across this https://www.youtube.com/watch?v=SfEZSyvbj2w
And everything seems to be clear to me how it is implemented, but one little thing is not clear. They have 64 action slots, as I understand the actions are done in order and then I had questions:
1. If everything in the world is random, and it does everything according to the list, will it not learn to survive, or how are the commands executed?
2. When the commands end, does the list start again? if so, then I don't understand how they still "learn" because the same actions in the same order
3. Are there conditions between the conditions? For example, he has a command "look there" this command, in theory, can return a new command to us, depending on what is "there" or it will violate the principle of the genetic algorithm.

Answer 1

https://svyatoslav.biz/education/testing_career_path/
https://svyatoslav.biz/software_testing_book/
https://github.com/anas-qa/Quality-Assurance-Road-Map

Answer 2

How can we act?

Working on connections, courses with practice (and / or possible employment), literature on preparation for professional certification .
146% will need English Intermediate+.
And no enikeystvo or technical support! Even a combination! Only specialized experience is needed.

Answer 3

Perhaps you can feed something else from here)
https://m.habr.com/ru/post/428690/

Answer 4

Once I asked the same question, only addressed to me, and I asked it on a toaster. Here is the link https://toster.ru/q/118099, plus a selection of his answers to the qa topic https://toster.ru/user/azShoo/tag_answers?tag=%D1%... , I highly recommend reading his advice, azShoo reale is a very cool dude. And so there is a channel in the telegram for qa juniors, it’s worth asking this question there https://t.me/qajuniors

Answer 5

The best option is paid or free courses with employment.
If the company is good, then they also pump over the technical part. I went through this myself, I've been working for a year

Answer 6

1) yes, randomly, but theoretically there is such an algorithm (sequence of commands) that allows the organism to survive as long as possible in this random world. Well, for example: if you make a sequence of moves such that the trajectory of movement will cover as large an area as possible, and dilute it with operations of eating from different sides, then the chances of an organism to survive are much higher than those of an organism that walks in one direction and expects food only from the left.
2) Yes, the command sequence is looped. The fact is that learning is not produced by an organism, but ... by a program that selects those organisms that have survived. A small part of the selected and propagated organisms undergoes mutation, i.e. their command sequence changes. if the change is successful, then this organism is likely to fall into the next generation, and there will be more organisms with such a mutation. By repeating this many times, you can find the optimal bot behavior algorithm. Most likely it will correlate with the world generation algorithm.
3) genetic programming is... not a specific algorithm, it's more of an idea that has steps that must be followed, but each step can be implemented in completely different ways. The conditions and commands here can be anything, this is just a description of the environment in which the optimization is performed.