what's your state? Is it just 2^number_of_cell on the tetris board? Like what is your formalized question? Given a shape that's your board, which next state reachable by an action is most good? I guess you can use more efficient state by only remembering the top siluet of the board? I'm curious ;)
did you train yours GAs using a specific static sequence of tetris blocks, or will it alwways perform this good with a random block sequence each time you run it? (without having to train it EVERY TIME)?
If it works with a random sequence then well done. I think GAs and NNs are the future and I fully understand the urge to show to ppl the real POWER of this algorithms!
Very beautiful indeed, good job! You mentioned the 5 value genome - what does the genome map to? I guess what I'm trying to find out is the genotype - phenotype mapping.
This is like about top 3 in the most beautiful things I have ever seen. Or, in Flynn's words... "Bio-digital jaazzz, maannn!" This week I saw glassesless 3D augmented reality, now this?? In 2010 I was disappointed we didn't make the future, now I'm like fuuuuck we're here! By the way is there any limit to how much a GA will learn, or a drop-off in learning over more generations?
@Sandcat87 I think that everything has a limit :) The limit of this AI, if we consider a very long training time, could be the resolution of the model of the world the AI uses to take decisions.
Simply above a certain generations limit It doesn't improve significantly
Hello, nice vid and ai, if you have statistical skills you could analyze statistical properties of your AI game for example series of heights of the highest filled grid or number of cells on the screen (each tetris block consist from 4 cells) binary series where 0 - no cleared row, 1 - cleared row, are you have you code available somewhere on the internet ?
hi, enjoyed your video, i would be interesting to see how it performs against bastet (barstard tetris) which is available for linux, and deliberately gives the player the most awkward shapes. if it can beat this im very impressed !
cheating, the ai seem to be running inside the game, you should place it outside and allways force start the pices at center just like in normal tetris then have it to move it accordingly into best fit possible location
Nice piece of work! I have also developer a Tetris AI (first in Java and then optimized in C++). It makes 1,800,000 lines in average when not knowing the next piece (11,000,000 is the record), and has made 100,000,000 lines in "normal mode" (aware of the next piece).
Nice work!
I'm doing a tutorial on how to implement this in Java.
Check it out: bitsrandomicos dot blogspot dot com
=)
fndTenorio 3 weeks ago
Hello there skynet
Agorsan 4 months ago
Hi, amazing work!
How does your board evaluation function works?
fndmiranda 5 months ago
Mind open sourcing it? :D
TheBlueShamen 5 months ago
how do we know its not actually a chinese nerd guy playing? haha, just kiddin, great work man :P
R4MIN4T0R 5 months ago
@R4MIN4T0R LOL
HeDo88TH 5 months ago
@R4MIN4T0R :-)))))))))))))))
jakubkrcma 3 hours ago
Comment removed
IcySkeptic 6 months ago
excelente!!!
Blackennew 7 months ago
what's your state? Is it just 2^number_of_cell on the tetris board? Like what is your formalized question? Given a shape that's your board, which next state reachable by an action is most good? I guess you can use more efficient state by only remembering the top siluet of the board? I'm curious ;)
dewinmoonl 7 months ago
I want to see Genetic Programming vs BATTLETOADS
fstratzero 8 months ago
Hmmm.....is this possible for more complex programs like car simulators??
waddlerobloxxxx 8 months ago
@waddlerobloxxxx It would be a lot more difficult, but i think that after a good problem formalization everything can be done with GA :)
HeDo88TH 8 months ago
did you train yours GAs using a specific static sequence of tetris blocks, or will it alwways perform this good with a random block sequence each time you run it? (without having to train it EVERY TIME)?
If it works with a random sequence then well done. I think GAs and NNs are the future and I fully understand the urge to show to ppl the real POWER of this algorithms!
keep it up dude
vafles 9 months ago
@vafles It was trained with pure random sequences :)
HeDo88TH 9 months ago
This is so beatiful, gratz man :-)
mamdoma 10 months ago
Very beautiful indeed, good job! You mentioned the 5 value genome - what does the genome map to? I guess what I'm trying to find out is the genotype - phenotype mapping.
Thanks for the upload, have a good day.
BlondBomber106 10 months ago
This is like about top 3 in the most beautiful things I have ever seen. Or, in Flynn's words... "Bio-digital jaazzz, maannn!" This week I saw glassesless 3D augmented reality, now this?? In 2010 I was disappointed we didn't make the future, now I'm like fuuuuck we're here! By the way is there any limit to how much a GA will learn, or a drop-off in learning over more generations?
Sandcat87 11 months ago
@Sandcat87 I think that everything has a limit :) The limit of this AI, if we consider a very long training time, could be the resolution of the model of the world the AI uses to take decisions.
Simply above a certain generations limit It doesn't improve significantly
HeDo88TH 11 months ago
Really impressive.
kid29a 1 year ago
200 lines? Wow, this is really good :)
theonlinegamers 1 year ago
This is amazing! Do you have the source code anywhere? I would love to look at it.
How complex a puzzle do you think ANN could be trained to play? I have some puzzles in mind but I don't know how good results I should expect.
RAWRimaTroll 1 year ago
Hello, nice vid and ai, if you have statistical skills you could analyze statistical properties of your AI game for example series of heights of the highest filled grid or number of cells on the screen (each tetris block consist from 4 cells) binary series where 0 - no cleared row, 1 - cleared row, are you have you code available somewhere on the internet ?
plmqas 1 year ago
you ai programmers should join harddrop.com its a pretty intense tetris community
Someone2knoe 1 year ago
What you call this warm up!!! She looks impressive.
cakegy 1 year ago
hi, enjoyed your video, i would be interesting to see how it performs against bastet (barstard tetris) which is available for linux, and deliberately gives the player the most awkward shapes. if it can beat this im very impressed !
infoburp 1 year ago 15
cheating, the ai seem to be running inside the game, you should place it outside and allways force start the pices at center just like in normal tetris then have it to move it accordingly into best fit possible location
FusionNinjin 2 years ago
i should have placed it outside, but i noticed that the "thinking time" was so short that there is no point in doing an "external ai".
HeDo88TH 2 years ago
ok, i tested the thinking time, it's about 3 and 3.5 ms.
external ai should have been an overkill :)
HeDo88TH 2 years ago 6
bella hedooo
artmar1 2 years ago
Search for "tetris ai" it's the one with the name TetrisAnalyzer.
tengstrand 2 years ago
ok i found it, very nice :)
i have just made a new version and i'll post it soon
HeDo88TH 2 years ago
Nice piece of work! I have also developer a Tetris AI (first in Java and then optimized in C++). It makes 1,800,000 lines in average when not knowing the next piece (11,000,000 is the record), and has made 100,000,000 lines in "normal mode" (aware of the next piece).
tengstrand 2 years ago
wow! can you post a video?
HeDo88TH 2 years ago
Folleeeeeeee xD (Michele)
Maikon90bBoy 2 years ago