AI Teaches Itself to Play Flappy Bird - Using NEAT Python!

Cant wait for the project tutorial! 😍

Wow, this is just awesome. Code bullet's videos are pure entertainment

Yours is worth a gazillion more from a learning perspective. Plus I actually find this very entertaining too.

As someone who has learned a bit of python for work even though it's outside my field, but has become interested in machine learning this is really fantastic. ML is super interesting to me and NEAT is wild. I've run the source code and it feels like magic so I've really gotta dive in and try to get a handle on whats going on. Classes are still something I'm not confident with lol. You've certainly earned a like and a subscriber though. Thank you for the video.

Tim, you sir deserve a LOVE not just a like. Thanks man you're a genius, and I've learned a lot from you. Keep Going bro !!!

Did u see the end

Loved the way you taught this entire tutorial. You taught in such a simple method, Really appreciate your efforts man! Looking forward to more tutorials!

Another rip off?

how many generations til all 100 go infinitely

Explained really well :) :)
Thankyou..!

Hi Tim
I wonder if this NEAT-python AI-programming adaptable for numbers and statistic, able to find patterns in massive data of numbers and combinations? Thanks a lot for your great content..

thanks!

one of the only good explaination channels iv seen

You are simply awesome!!!

Very Good work done

I'm NEET too.

You should do this kinda intro videos ... it will bring us to your playlist 💕💕💕

Hey Tim, Your tutorial series is amazing! Learnt a lot. Continue making more and more tutorials!!

Thank you Tim, i really enjoy learning python by watching your videos!

Excellent explanation on AI using a simple example, No Wonder Microsoft hired you. Keep up the great work. Love from India

Can you tell me why you chose TanH instead of Sigmoid activation function?

Yeah I want more! :D

thx for the video , i was hoping that it's more than one Perceptron connection ...... i think that if you put a victor or a ray from the eye's and let it see as it's go's up and down where the gap is, it will be smarter rather than given it where the gap is. thanks again

So in the beginning you give random weights for the starter population right. But in what range do you choose those starter weights?
Also, usually when i see video's like this one there is also a hidden layer in the network. What is the benefit of that hidden layer, if it works without one like you did?
EDIT: i tried using starter weights and bias of random value between -0.5 and 0.5, but the birds didn't learn anything. I dont really know if i should do something with my distances before passing them to the weighted sum? I get like random values of -200 to 200 for distances

what about playing browser game by reading screen with cv2 instead of python game

This video is awesome. Thanks Tim for all your content. Keep going!

Incredibly helpful, thanks so much

I'm mechanical engineer but I'm just curious about how it works.... Good work bro

Can you make the game harder for the A.I. And almost impossible for human play. And see if it can still win. Rand pipe Len, pipe width, and rand pipe distance between the next set of pipes. A long with pipe tilt... basically chaos. With such precision required that only a machine could do it.
. 0. O
O. O
O. O
O. O. O
O. O. O
O O. O
O. O. O
O. O
O. O. O
O. O
O. O. O
O. O. O
O. O. O
O. 0. O
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

Love this style of video!! Keep doing whatever you want to do. You'll always do it great!!

How do you interface neat python code with flappy bird game?

Since tanh(x) is almost linear in the [-0.5, 0.5] interval, does it make any difference here to use an activation function ?

thank you

New sub this was so helpful I have never seen someone explain it in a way i can understand AND give code AND like a source
thank you so much!
Edit: AND THE MODULER????

😊♥

Thanks bro! i genuinely appreciate your hard work on this video!

Did you have to create the flappy bird yourself? Or does the AI play the actual game on phone?

nice

This is crazy

Good effort Tim, well done!

I liked the style of the video, but some code-example would have helped to understand how to implement NEAT into a game

I know i am late but please make every video u make like this one. Mind blowing

Amazing video. I'm curious about what made you pick "> than 0.5" as a threshold to jump? Could it be 0 or any other numbers (like -0.5 or -0.7). Thank you so much.

nice video bro

I can’t get it to work on my own system or with replit. It says no module “neat” found even though I’ve used pip install neat

I am going to try to make this as well, thanks for the info. I would love to see a video of how you went about coding it.

love this one!

good video