Hi. My name is Abishek Nandy. I’m here at SIGGRAPH
showcasing my work, UnityML, using the Intel Optimized
Python Pipeline. [MUSIC PLAYING] This project is
about Unity machine learning bringing
in the power of CPU to more optimize
your simulations. This solution is based
completely on CPU orientation. You can see on
the demonstration, on the left-hand side
it’s an untrained model. We see that the balls
are falling off. There is a plate. It’s trying to support it. But after 50,000 iterations on
the CPU, we get this output. It’s trying to learn. It’s balancing on its own. We have used the Intel Optimized
Python as well as a Tensorflow pipeline for our project. We are trying to
optimize it for the CPU so that, if we have a
normal desktop at our home, and if we want to build so
many applications on our own, we can give it a try. Like the researchers who
are working on Unity machine learning, if they have some
research areas, research papers, it’s very easy
and cost-effective to bring your projects to life. So we can do the simulations
here using the Unity, as well as the Intel
Optimized Python Pipeline. And we can integrate it, and we
have a better solution for it. We see that there
are a lot of changes happening in machine learning. And you see that there
is a difference in terms of shifting towards
reinforcement learning, now more towards
imitation learning. We are doing an integration
with virtual reality, too. What we are trying
to do is we are trying to build up many
applications using the Unity pipeline. And we have different
areas, like if you want to support different
kind of simulations, like object creations, their
movements, learning on its own, you can definitely give it
a try using reinforcement learning in Unity. About my projects,
you can definitely give a try at
devmesh.intel.com, where I share my details and my work. And about the program, like
the Intel Software Innovator, you can definitely check
on it, software.intel.com. Thank you.

UnityML Theory Using Intel® Optimized Python Pipeline | Intel Software

2 thoughts on “UnityML Theory Using Intel® Optimized Python Pipeline | Intel Software

  • September 24, 2018 at 7:39 pm
    Permalink

    awesome sir

    Reply

Leave a Reply

Your email address will not be published.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