Deep Learning for Robotics - Prof. Pieter Abbeel - NIPS 2017 Keynote

Поделиться
HTML-код
  • Опубликовано: 3 окт 2024
  • Pieter Abbeel delivers his keynote: Deep Learning for Robotics, at NIPS 2017.

Комментарии • 18

  • @RobertsMrtn
    @RobertsMrtn 6 лет назад +4

    One of the best videos on machine learning.

  • @TheAIEpiphany
    @TheAIEpiphany 3 года назад

    Now that was dense haha! Thank you for the talk!

  • @brucesmitherson3545
    @brucesmitherson3545 6 лет назад +5

    This guy knows more than Andrew Ng

  • @suvirmisra2730
    @suvirmisra2730 6 лет назад

    We are just there to get intelligent robots that will learn to work around in 3 d spaces easily on its own.

  • @junqingqiao8007
    @junqingqiao8007 5 лет назад

    Really great video!

  • @squarerootof2
    @squarerootof2 6 лет назад +1

    Was that maze Wolfstein?

    • @autohmae
      @autohmae 6 лет назад

      Well Wolfenstein 3D source code was released for free in 1995

  • @lilyz.4218
    @lilyz.4218 5 лет назад

    leverage simultaion 33:04

  • @stephenstreet1045
    @stephenstreet1045 6 лет назад

    Can anyone please recommend some quality videos on this topic pitched at a more rudimentary level?

    • @jomen112
      @jomen112 6 лет назад +1

      ruclips.net/video/ELn-1iP3BDs/видео.html

  • @123darkdeal
    @123darkdeal 6 лет назад

    I want to know about AI and the mastery of language (words, definitions, and grammar). Fully understanding what words mean. From there, would it be possible for an AI, with a full vocabulary, to formulate a hypothesis and do it's own research/experiments to learn from there. It seems that all AI are learning from basic trial and error, and that is time consuming.
    Basically, it would be neat to get an AI to the point where it doesn't have to be programmed for each task. You could ask it something and it would figure it out on its own. Even exceedingly complex questions that people don't have the answers for yet.

    • @shacharh5470
      @shacharh5470 6 лет назад

      If your goal is an AI that can formulate and test hypotheses, why do you start from natural language? In many domains hypotheses can be expressed formally so there's no need to use natural language

    • @jomen112
      @jomen112 6 лет назад

      me to me to me to

    • @borilboyanov5544
      @borilboyanov5544 6 лет назад

      "You could ask it something and it would figure it out on its own. "
      WARNING! 42. You know the value of the universe and all it means and the fate of the planet that started it... i guess you know that...

  • @CandidDate
    @CandidDate 6 лет назад +15

    Once we get a fully functional humanoid robot, we can mass produce them and send them to Mars or the Moon to initiate colonization. Call up Elon Musk! This is a wonderful time to be alive for sure!

    • @TheHeavenman88
      @TheHeavenman88 6 лет назад +4

      Elon is way ahead of you. This reseach is done at Open Ai which is owned by Elon Musk

    • @autohmae
      @autohmae 6 лет назад

      Not the research of Pieter Abbeel, his research is done at Berkeley.

    • @damienstanton
      @damienstanton 6 лет назад +1

      He was one of the cofounders of OpenAI, but he certainly does not own it. He is no longer even on the board. OpenAI is a nonprofit research organization.