Bharath Ramsundar
Bharath Ramsundar
  • Відео 24
  • Переглядів 21 776
AlphaGo Zero: AI of the Year?
Bharath Ramsundar and Patrick Hop talk about AlphaGo Zero, Google's new self-taught system for Go play that beats the best human masters without any human training. What does AlphaGo Zero mean for AI and the advent of artificial general intelligence.
Переглядів: 1 563

Відео

Is AGI something to worry about?
Переглядів 5214 роки тому
Bharath Ramsundar and Patrick Hop battle it out about whether the singularity is going to kill us all! Or more soberly put, is artificial general intelligence likely to happen in the near future?
DeepChem Meetup 9/22/2017
Переглядів 6894 роки тому
DeepChem User Group Meetup on 9/22.
DeepChem Meeetup 8/9 Continued
Переглядів 2744 роки тому
DeepChem Meeetup 8/9 Continued
DeepChem Meetup 8/9/2017
Переглядів 5944 роки тому
DeepChem User Group Meetup on 8/9
Livestream Test
Переглядів 604 роки тому
Testing livestreaming
The Rational Exuberance of Deep Learning
Переглядів 2,9 тис.5 років тому
The Rational Exuberance of Deep Learning: The breakthroughs in imaging and natural languages can be replicated in chemoinformatics (Vijay Pande)
The Thicket of Challenges in GPCR Molecular Pharmacology
Переглядів 6005 років тому
The Thicket of Challenges in GPCR Molecular Pharmacology: Paradigm shifts bring new opportunities and new hurdles (Ryan Strachan, UNC Chapel Hill)
The Holographic Pose
Переглядів 6035 років тому
The Holographic Pose: Docked conformations as Tensor Images (Abraham Heifets, Atomwise)
Chemoinformatics for Pharmacology
Переглядів 1,4 тис.5 років тому
Chemoinformatics for Pharmacology: Why it shouldn't work, why it does work, and opportunities for predicting new biology (Brian Shoichet)
Democratizing Drug Discovery with DeepChem
Переглядів 3,4 тис.5 років тому
Democratizing Drug Discovery with DeepChem (deepchem.io). (Bharath Ramsundar)
Virtual Screening Step by Step
Переглядів 9 тис.5 років тому
Virtual Screening Step-by-Step: Combining computational methods and expert judgement (Magdalena Korczynska, Shoichet Group)
Virtual Screening Step by Step
Переглядів 9 тис.5 років тому
Virtual Screening Step-by-Step: Combining computational methods and expert judgement (Magdalena Korczynska, Shoichet Group)
Democratizing Drug Discovery with DeepChem
Переглядів 3,4 тис.5 років тому
Democratizing Drug Discovery with DeepChem (deepchem.io). (Bharath Ramsundar)
The Rational Exuberance of Deep Learning
Переглядів 2,9 тис.5 років тому
The Rational Exuberance of Deep Learning: The breakthroughs in imaging and natural languages can be replicated in chemoinformatics (Vijay Pande)
AlphaGo Zero: AI of the Year?
Переглядів 1,6 тис.4 роки тому
Bharath Ramsundar and Patrick Hop talk about AlphaGo Zero, Google's new self-taught system for Go play that beats the best human masters without any human training. What does AlphaGo Zero mean for AI and the advent of artificial general intelligence.
Chemoinformatics for Pharmacology
Переглядів 1,4 тис.5 років тому
Chemoinformatics for Pharmacology: Why it shouldn't work, why it does work, and opportunities for predicting new biology (Brian Shoichet)
DeepChem Meetup 9/22/2017
Переглядів 6894 роки тому
DeepChem User Group Meetup on 9/22.
The Holographic Pose
Переглядів 6035 років тому
The Holographic Pose: Docked conformations as Tensor Images (Abraham Heifets, Atomwise)
The Thicket of Challenges in GPCR Molecular Pharmacology
Переглядів 6005 років тому
The Thicket of Challenges in GPCR Molecular Pharmacology: Paradigm shifts bring new opportunities and new hurdles (Ryan Strachan, UNC Chapel Hill)
DeepChem Meetup 8/9/2017
Переглядів 5944 роки тому
DeepChem User Group Meetup on 8/9
Is AGI something to worry about?
Переглядів 5214 роки тому
Bharath Ramsundar and Patrick Hop battle it out about whether the singularity is going to kill us all! Or more soberly put, is artificial general intelligence likely to happen in the near future?
DeepChem Meeetup 8/9 Continued
Переглядів 2744 роки тому
DeepChem Meeetup 8/9 Continued
Livestream Test
Переглядів 604 роки тому
Testing livestreaming

КОМЕНТАРІ

  • SH` Bangash
    SH` Bangash Рік тому

    what is the related virtual screening methods ? and kindly upload the video about If the 3D structure of a target is known, which methods could be used to do virtual screening? If no crystal structure of a protein-ligand complex is available, but at least one known ligand of this target, which methods could be used to do virtual screening?

  • ROOPAL GUPTA
    ROOPAL GUPTA Рік тому

    really good and informative

  • Yerdos Ordabayev
    Yerdos Ordabayev 4 роки тому

    I will be worried the day AI rage quits during the game.

  • p3rsi4n1
    p3rsi4n1 4 роки тому

    So poker is much deeper game then Start Craft or Dota or Go for one simple reason, it adds the element of Bluff. In order to figure out if your opponent is bluffing, players have to "get a read" on there opponents which is anything from game play history, to play style all they way to micro hand or facial movements. To make an AI that's at a pro level poker player, it would have to be able to read human emotions through game play and even through visual cues. This I believe we are still far from.

  • Renaissance Man
    Renaissance Man 4 роки тому

    You're talking about technology, but you recorded this on a phone from 2005. Come on bruh

    • p3rsi4n1
      p3rsi4n1 4 роки тому

      agree, I found the talk interesting but the audio and video quality is very poor...

    • Renaissance Man
      Renaissance Man 4 роки тому

      I can't understand 50% of the convo, very good Intel, but very poor quality

  • Fabian Schreiber
    Fabian Schreiber 4 роки тому

    Thanks for recording/sharing. The talks start at 18:48

  • Connor Davis
    Connor Davis 4 роки тому

    Some updated stats - I didn't realize the nVidia "Tesla" K80 and M60 cards have Kepler and Maxwell chipsets respectively (not just Tesla). Here are my updated cost estimate (both including and excluding est. CPU PPD production): Cost per 1 PetaFLOPs/day on AWS instances using 3 year up-front contracts: Instance GPU Cost 1 PFLOP/Day Num Instances Req. g2.2xlarge (est GPU + CPU est) 1 core Kepler GK104 $2,513.46 371 p2.xlarge (est GPU + CPU est) 1 core Kepler GK210 $2,489.55 237 g3.4xlarge (est GPU + CPU est) 1 core Maxwell GK204 $2,425.07 200 ----------------------------------------------------------------------------------------------------------------------------------------------------------------- g2.2xlarge (GPU Only) 1 core Kepler GK104 $2,957.01 437 p2.xlarge (GPU Only) 1 core Kepler GK210 $2,676.94 279 g3.4xlarge (GPU Only) 1 core Maxwell GK204 $2,853.03 235 Looks like the *g3.4xlarge* instance is currently the most cost effective based on a 3-year contract Sources: en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units calculator.s3.amazonaws.com/index.html aws.amazon.com/ec2/instance-types/ (using Accelerated Computing option) Edit: GK204 is an nVidia Maxwell chipset _Note: For performance calculations, used average of standard and "Boost" performance_

    • Connor Davis
      Connor Davis 4 роки тому

      Remember, when AWS says you're getting a K80, or M60 ... you are only getting ONE core of that card NOT both in a single GPU instance ;-)

  • Connor Davis
    Connor Davis 4 роки тому

    At 3:26, there must be a miscalculation. To get 1 PFLOP on AWS using the most cost-efficient Folding@home compatible GPU instances (Kepler based) - it would take 474 separate g2.2xlarge (nVidia GK104) EC2 instances under a 3 year reserved contract to generate 1 PFLOP of computational power on a daily basis. This would cost a user about *$3200 per day* _NOT $100 per day_ . Using on-demand pricing would be even more astronomical! Not even the lowest spot instance pricing could get to $100/day for one PFLOP on cloud GPUs. Please correct me if I'm wrong. Here is the official calculator for Amazon Elastic Compute Cloud (Amazon EC2) pricing. calculator.s3.amazonaws.com/index.html Thanks. _Edit: Updated nVidia model and total number of required EC2 instances to reach 1 PFLOPS per day_

  • Dan Elton
    Dan Elton 4 роки тому

    Did you mean to post this? This looks like the camera was just running during the A/V setup....

    • Bharath Ramsundar
      Bharath Ramsundar 4 роки тому

      We had to interrupt the live-stream so this got posted (recording during setup). Will probably remove this video once I have some cleanup time.

  • Dan Elton
    Dan Elton 4 роки тому

    Great!! Count me on board. I am currently playing with the molecular autoencoder in my research and would love to link it with DeepChem to do hill climbing / Bayesian optimization in chemical space. Due to the low-ish resolution of the video, I would love to see a copy of the slides!

    • Fabian Schreiber
      Fabian Schreiber 4 роки тому

      Slides are here: deepchemworkshop.docking.org/slides/Ramsunder.pptx

  • Mostapha Benhenda
    Mostapha Benhenda 5 років тому

    where are the slides?

    • Fabian Schreiber
      Fabian Schreiber 4 роки тому

      slides are here: deepchemworkshop.docking.org/slides/Ramsunder.pptx (sorry for the repeating posts)

  • Miao Wang
    Miao Wang 5 років тому

    Hi, could you provide the PPT file ? Thanks.

    • Fabian Schreiber
      Fabian Schreiber 4 роки тому

      slides are here: deepchemworkshop.docking.org/slides/Ramsunder.pptx