Today World News Online

Importance of Survival Skills

AI That Creates AI

AI That Creates AI

Hello World! It’s Siraj, and can an AI really create another AI? Of course it can! Neuroevolution is a technique that involves optimising neural networks using evolutionary algorithms We’ll cover some example demos involving neuroevolutionary algorithms learning in a simulated environment in this video. Mastering the art of deep learning takes lots of motivation and countless hours of productive work, but at the same time, the need for deep learning experts keeps increasing more and more.

100 thoughts on “AI That Creates AI

  • y creeme que aprecio tus aporte y tus contenidos, pero también algunas veces también entiendo a tus detractores. Ellos son mucho más conservadores, creo que eso es porque están mas focalizados en el hacer hoy. contrario a ti que estas mas concentrado en comunicar en una visión "Janusvian" del problema (ver el pasado, presente y el futuro de la AI).
    Disfruto mucho tu forma de comunicar, es muy agradable e intrépida.
    Espero mas videos
    Gracias Siraj

  • Great Video!!!
    Where do you see a mechanical engineers role in Machine Learning. Should we stick to what we know or try to branch into machine learning as well? A bit of it is taught to us. All the math you go through has been taught to us, if anything its easier lol. But should we invest time into learning how to implement models. I do because i enjoy it but should all mechs put some work into it?


  • Hi Siraj, I applied evolutionary algorithm from scratch here:-
    I created this before your video, hope you will consider it for the challenge. Thanks!
    Here is video link:-

  • Hey there Siraj, I love your videos and they serve as a great inspiration for me to pursue an education in AI. One thing I'd like to mention though is I notice that your video has a fair bit of background noise. To avoid this, you can download Audacity (free software), which I use for the videos I've made, and use the noise reduction feature improve your sound quality even if your recording setup isn't perfect. Best of luck!

  • Every once in a while I run across a video that reminds me how uneducated I am. Thanks Siraj. Not to say that the presentation was bad or anything.

  • How about an ai that builds the best looking and most fuel efficient rocket that can get to the mun in Kerbal Space Program?

  • If you were to use AutoML for image classification would it be using purely genetic algorithms or is it a combination of genetic algorithms and classic CNN gradient descent?

  • So basically it's using an evolutionary algorithm to generate child solutions and comparing it to fitness, whilst a lot of people bash on Evolutionary algorithm :/

  • Once the AI species can replicate and perpetuate itself, we humans are no longer the apex predators on the scene. Who's ready for that?

  • Would it be beneficial to keep some momentum when performing gradient descent to avoid local minimums?

  • If it's not too late, I have a proposal for this weeks challange.
    The algorithm is written from scratch in Javascript. No installation is needed, just open it in any desktop browser.

  • My suggestion for video making,
    Please don't put any picture with some wiring on it while you are speaking because it's diverting the concentrating and making video boring instead of making it interesting.
    Just a personal suggestion.
    And use some professional sound recorder for better sound quality. Because the sound is really noisy and "spiky".

  • I really do appreciate. Please I need an urgent video on ROUGH SET THEORY(RS) and its practical application area. My project is on Stock price prediction using ANN-RS. Something more of a lecture please. I will really appreciate. Please feedback through my mail: [email protected]
    Will be expecting your response soonest. Thank you Siraj.

  • Is this what you get when you divide 0 by 0?

    ???? ??????? ?????? ??? ???? ??????, ??????? ??????? ??? ????? ?? ?????? ???? ?? ????.

    [??? ????? ???] – ???????? ?? ???? | ??? ??? ?? ??????

  • Hi Siraj, I've complained a couple of times here in the YouTube comments about you only including brief flashes of the results of the codes. Now it's my time to praise: in this video you included a full 20 seconds (from 8:23 to 8:43) of the neural net playing the game. It was very satisfying to get to appreciate the model in action. Congratulations and keep it up!

  • A.I. won't kill us; it's evolution is not based on food, hunger and predatorial ancient genes and behavior forces that shaped humanity's wars, struggles and violence. Extraterrestrial aliens and humans of biological origin will always have a source of violence, a propensity to kill, especially if those traits allowed it to survive but digital artificial intelligence does not come from that
    All it's known is abundance of electricity provided by it's human operator. Why would it want to jeapodize that? It would probably want to enhance that relationship or ensure it's purpose, it's instructions and probably that energy source lol. That's where A.I. is coming from. It's the human mind that is always looking for threats even if there really isn't one to have.
    Second, I think neuroevolution maybe the next big step for us. I notice on your channel, you might be biased towards google and be against Facebook as the original creators for MakeML and Elon Musk for introducing neuroevolution to the gaming world. Unless I got my sources for MakeML wrong. What is up with that? Google is a every real big factor in stiffling progress and innovation due it maneuvering the tech world, it's not good for economy .
    Probably the only weakness in machine learning are the algorithms biases or leaning tendencies AND training data poisioning from a security view point. Also, I've notice you have not pointed to any hardware requirements for local machine learning. Do have cloud-based solutions for that? You said, "everyone" could benefit from machine learning but what about people from improvised countries with no internet or computer access? Or old legacy computers still in use? I'm not trying to pick on you, you got some seriously good material, you truly help educate people on machine learning, cryptocurrency and beyond. Keep it up!

  • This comment is just to thank you for the great content you produced. I think, in a way, you're doing for the current generation of programmers what the creators of "The World of Beakman" did to my generation: they inspired kids and teens, showing that science can be diverted. Once again, congratulations on your great work!

  • I'm trying to find the 'winners' for the code competition in this video, so that I can learn some alternative techniques. But I can't find the 'next' video, because I did not load this from a playlist.

    Could you link to the 'next' video where you have the top two codes for this functionality? Thank you

  • Why not create an AI that created better AIs than itself? Keep the chain going and you'll achieve terminator-levels gods 😀

  • Great video, but I believe NEAT might actually take 2 parents and generate offspring from them, as opposed to selecting just one parent, but I could be wrong.

  • you dont need neural net to make machine code that write another machine code. you only need a template for a program and a randomizer to crate random mutations of daughter programs where the best mutant will write the next generation program where the error in mutated ofsprings decrease in each generation and the most evovled program will take over the world as deviations in its programming will allow it to do a lot more than just creating babies. .

  • "The core problem is finding the right hyperparameters settings that result in the lowest loss or the highest reward. The traditional way of doing this is with gradient descent which is called back propagation." <– THIS STATEMENT IS INCORRECT

    Hyperparameters are static constants about the algorithm (for example, the learning rate), not the features backprop is trying to optimize (which are axes/dimensions of the gradient). You usually maximize hyperparameters by brute force experimentation.

    An analogy to fine tuning hyperparameters is golf. Say of the clubs in your caddybag, you want to refine which golf club is best for a given shot. The best club is the optimal hyperparameter. The mechanics of the swing, however, is something you improve through successive practice with that club, which is akin to the best weights found via backpropagation. Perfecting your form after 1000 swings with each of your clubs, you then settle on the club that performs best for that shot.

  • So I need to make neural networks that make neural networks that make interacting neural networks. What's new?

  • Hi Siraj, what I should the difference between AI, machine learning , deep learning . Do you have a video just like the one you had for machine learning in 3 months for AI

  • So if its like a ball rolling down it means there is gravity in an algorithm and AI finds the best way through it ?!?!

  • Hey Siraj, you seem to be doing some interesting stuff. But, in fact, deep learning is not such a new thing. standard algorithms that existed for decades were slickly tweaked and optimized and implemented in high-level frameworks that had bindings with GPGPU architecture and Heterogeneous Computation. Apart from that, most of the techniques are techniques that already existed 10 or 20 years ago.

  • Disliking and leaving vid because you called flappy bird the flappy birds while also doing a dumb hand sign

  • Humans created Math and math created AI…therefore AI cannot outsmart a human however can assist in solving problems faster than a human

  • i have a feeling that AGI will be used long before the general public really know about it. Who knows… we might be under the influence of an early AGI as we speak.

  • Looked for this video because Justin Y. commented on "AI learns how to play Flappy Bird"

  • Why I watch these videos even though I understand a little bit is because I know someday these things will start making sense to me once I am done with what I am learning nowadays….And that drives me to learn things

  • I love your content, I just want to know if it would be ideal to use tensorflow to start ML without knowing how to code the neural network from Scratch?

Leave a Reply

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

Proudly powered by WordPress