Small Basic AI Championship

[AbsoluteBeginner]

Z-S,
the most important question is whether this is allowed to be done.
If I have the right to place data in the source code of the executable file, then this will not be difficult to do. 

[z-s]

why it will not be allowed.
in my opinion it should be clearly allowed.

[AbsoluteBeginner]

why it will not be allowed.
in my opinion it should be clearly allowed.

(translated by Google translator)

I'm no expert, but I've heard that it's not recommended to combine executable code and data in the same file. 

[litdev]

AB,

I think you can put any text in a file you publish/import.

If its not a large amount of data this would be fine I think, but probably not what was intended.

It is generally best to store additional data in files rather than source code so you can change the data without recompiling a program, but again whatever works best in your case is fine.

If you have a bunch of files, including source that users need to modify before they can use, you may be better using some file sharing site with all your files in a zip, like ZS did for his extension.

[AbsoluteBeginner]

(translated by Google translator)

Yes. And I think so.
The files are not large. Before compilation, the user will cut out the text with these data and simply save the text on his disk in a text file.
But I automate even this operation as much as possible for the user.

So, with your permission, I am training the second and third Rabbit mutations and publishing the codes along with the training results. 

Have a nice day everyone. 

[AbsoluteBeginner]

(translated by Google translator)

So, I published a new version of the third mutation "Rabbit AI": XBWP716.000-5

This is the luckier "Rabbit" of the two mutations that have some semblance of learning ability.
This “Rabbit” remembers and passes on to its descendants in genetic memory not only successful sequences of four actions, but also the places on the field where the “Rabbit” managed to survive most often.
When you first start the Client, the program suggests either using preset training data or starting training from scratch.

I declare this “Rabbit” to be my official cybersportsman at our “Small Basic AI Championship” in the game “AI Snake” for the category of participants who do not use neural networks. 

( if no one else presents their athletes, then my “Rabbit” will AUTOMATICALLY become the Champion!!! So don't try too hard.   )
- - - - - - - -

The rest of the programs in the game "AI Snake" also have fixes, but very small ones.

Program "Nature": RPSQ181.000-10
Client program "Snake AI": TJSG543.000-4

And now I’ll start working on a neural network for my new cyberathletes.

[z-s]

I Don't know what's happening when i started nature and connected rabbit and snake.
Snake is so powerful every time he knows rabbit location and attack him his size is increasing only and the rabbit is just jumping and getting caught by snake.
Snake When Go Through Rabbit His Size Increase But Rabbit Don't Die Could You Explain ME.

[AbsoluteBeginner]

I Don't know what's happening when i started nature and connected rabbit and snake.
Snake is so powerful every time he knows rabbit location and attack him his size is increasing only and the rabbit is just jumping and getting caught by snake.
Snake When Go Through Rabbit His Size Increase But Rabbit Don't Die Could You Explain ME.

Hello, Z-S.

Yes, the Snake always sees the Rabbit, and the Rabbit never sees the Snake.
Until we come up with new rules for the game, the Rabbit will not be able to see the Snake when the Snake is sneaking up, because the Snake has excellent camouflage.

Therefore, the Rabbit must perform unpredictable actions so that the Snake's head cannot end up in the same cell of the field where the Rabbit is. If the Rabbit and the Snake's head are in the same cell, then this means that the Snake has caught the Rabbit.

But the Rabbit can move where the Snake's body is.
Since we look at the field from ABOVE, we see that the Rabbit seems to be walking on the Snake. The Rabbit perceives the Snake, which is very well camouflaged, as a tree branch that lies in the grass. The Rabbit easily moves through the cells of the field on which the Snake's body lies.

When the Snake caught the Rabbit, this Rabbit dies.
The Snake's body becomes longer, and a NEW RABBIT appears on the field.
But, the Rabbit can defeat the Snake if the Snake falls into a trap.
A snake cannot crawl on its own body. Therefore, when the head of the Snake does not have the opportunity to move anywhere, then the Snake dies.

Therefore, in order to evaluate the results of the game, the ratio of the number of eaten Rabbits to the number of dead Snakes is used.
My last Rabbit allowed Snakes to average 13 segments long. This is much better than the average length of 20 segments that Snakes have at the beginning of Rabbit training.

[z-s]

I get it now thanks for explaining

[AbsoluteBeginner]

Hello friends. 

I can't choose which direction to continue my movement, developing a new mutation of Rabbit.
I mean Rabbit, which uses a neural network.
We know that neural networks can have various structures.
But, if you use a neural network in the game "AI Snake", then what rule should be followed?

Should I choose the neural network structure I want to use in the competition or should I create code that will generate random neural network structures?
A random structure generated by the code will of course be more interesting. But checking each of the random structure options will take a lot of time.
If a person develops and trains the structure, it will be faster, but not as interesting as a random process.

Tell me what will be more interesting to you personally?