11.1: Introduction to Neuroevolution – The Nature of Code

[ad_1]
Welcome to a new topic in the Nature of Code series: Neuroevolution!

🎥 Next Video:

🔗 Toy-Neural-Network-JS:
🔗 Nature of Code:

🎥 My Neural Networks series:
🎥 My Genetic Algorithm series:
🎥 My Doodle Classifier series:
🎥 Flappy Bird Coding Challenge:

🚂 The Coding Train website:
💖 Support this channel on Patreon:
To buy Coding Train merchandise:
📚 Book recommendations:

💻

🎥 For More Nature of Code videos:
🎥 For More Coding Challenges:

🔗
🔗


Posted

in

by

Tags:

Comments

21 responses to “11.1: Introduction to Neuroevolution – The Nature of Code”

  1. Shinrod Dellore Avatar

    Hey ! I'm so happy that you've started this series !
    I've tried to make an AI able to play snake based on your Genetic Algorithm series and the Neural Network one, but it has been a few weeks that I'm in trouble to make it work…
    I hope this will help me out.
    Btw, keep making your awsome videos 🙂 I understood more than I could hope thanks to your job !

  2. realcygnus Avatar

    exciting !…….this should get interesting

  3. Ahmar pengoo Avatar

    Plz make a video Q/A video on how you learn, and a little background of yours.

  4. Vincent Sim Avatar

    Yeaaaaahhhh ! :)))

  5. Hydrozoa Avatar

    You said that the neural network is the genotype, but it's actually the phenotype. The DNA (connections and nodes) is the genotype, from which we create the phenotype (the network). I have some videos on this exact topic.

  6. Flame Stryke ShadowDark Avatar

    I would suggest you just build a crossover function with 3 inputs, nn1, nn2, and crossover point "a", then have that function take the first "a" components of nn1 and the last length – "a" of nn2 and swap those to create nn3 and nn4. If you're just copying the neural network, you can simply pass the same network as both nn1 and nn2. It will save some refactoring in the future.
    Alternately, have only 2 inputs and have the function generate it's own crossover point, but if you pass in a crossover point you can have it cut the networks at the same place every time for debugging purposes.

  7. J0ooop Avatar

    Awesome timing! After watching your NN series and Genetic algo series, I started making a Snake simulation with Neuroevolution. It's almost done right now, still slightly tweaking my fitness function. Thank you so much for your work, your videos are great.

  8. Yousif Al-Yousifi Avatar

    11:00 you went full neural network…. never go full neural network

  9. ZakNikFMP Avatar

    it would be cool if you do some on this in python 😀

  10. That guy who commented Avatar

    Create a neural network that creates other neural networks, that would be really interesting im my opinion!

  11. Aasim Baig Avatar

    I would love to read The Nature of Code 2 …

  12. mike jones Avatar

    how about a "free will" learning algorithm? but then it'll be like "i dont want to look at cat pictures" so if we're just telling it what to do, then why dont we just tell it what to do? whats the point of learning? i suppose is my question, if theres a desired outcome, whats the learning for?

  13. AADITYA RANE Avatar

    Great….✌
    From last several months I was waiting for to learn this via coding train.

  14. Code Blooded Avatar

    Which company hire a processing and p5 expert??

  15. sanfrozen Avatar

    sir u make awesome video that taught new thing to me.keep it up the good work.

  16. Codie Petersen Avatar

    You should be careful with that fitness function. The broader the ruleset the longer it will take to converge on an appropriate answer. Instead of doing "how long I live", you could also maybe put in some other variables, like amount of movements, because ideally you would want something that would respond like a human, or how long the agents stay within the entire horizontal gap region gap.

    Also, it would be cool if you had a way to identify the different agent populations from one another. Just so you can see the specific generations propagating over time through each trial run.

  17. HM Game Avatar

    the dream is the coding train reinforcement learning video

  18. RIT123 Avatar

    Happy Great Ape Day folks!

Leave a Reply

Your email address will not be published. Required fields are marked *