Artificial Intelligence Conference 2018 – San Francisco, California
O'Reilly · 23 lecciones · Gratis en YouTube
Using machine learning in workload automation (sponsored by Digitate) - Akhilesh Tripathi (Digitate)
Sobre este curso
the second example I'd like to walk through comes from Hollywood how many of you have watched the movie the Meg OKC one all right you guys this is like the
best shark movie maybe after sharknado and Jaws okay so you really need to go see this if you haven't I bet you did not know that the prehistoric Megalodon that stars in the Meg was actually
rendered using machine learning so let's watch a video where Ziva dynamics and scanline visual effects talk us through how they created it when working on creatures the very important aspect if
if not the most important aspect is to create an underlying machinery that triggers all the different aspects of anatomy at the right time it's a very complex relation between all of these
different aspects the ways that we use Intel Xeon scalable processors is to generate amazing training data so like if you look at like all the complex motion that you see on the gills like
all the wave propagation that you see all that stuff is a product of running simulations using our software so y'all need to go see that again top
shark movie after sharknado I mean that's incredible so making the Meg real what James was saying from Ziva dynamics I mean what we did was learn from the movements of the
shark creating models that studied the anatomy as well as the flow that the shark took through the water to help create this incredibly photorealistic creature that's you know Hollywood scale
and if you look at the sheer amount of data that was processed we're talking about 2 million you know just millions and millions of 2d and 3d images being processed on tens of thousands of Intel
Xeon CPU cores and ultimately what that's delivering is savings in time we're able to see these photorealistic creatures created not so much in thousands of human hours but just like a
few machine hours you
SIEMPRE HAY ALGO MÁS