AI Luminary Series: Naveen Rao Discusses Deep Learning Tools for AI Advancement
Artificial Intelligence Luminary: Naveen Rao, VP of Machine Learning Solutions at Intel
In this video, Naveen Rao discusses how new deep learning tools and computing advancements in AI will enable new ways of thinking.
Read the transcript:
My definition of artificial intelligence is really the ability to find useful structure in data somewhat automatically. If you think about what a brain really does, we take in information from our senses, from our eyes or ears or touch, and we actually collapse that down into some useful structure that allows us to act on the world and survive.
Tiger’s coming at me I know to run, vegetables of a certain color or characteristic or something are good to eat. My background was really in computer engineering, but I always had an interest in biology and how biology solved computation. I know how we build a computer; what can we do better by looking at what biology has done.
4 billion years of evolution had to produce something interesting, so why is there a disconnect between what brains can do and what computers can do? And that’s kind of where we started with Nervana.
Nervana is a, full-stack deep learning AI platform. The comprehensive set of tools to allow data scientists to access latest greatest techniques for the data. Organized tools that actually break things down and make it simpler. And that’s exactly what’s lacking right now in the AI community.
The hardware we have today is not fundamentally purpose-built for this problem and has lots of constraints and restrictions, and you see that in the algorithms and even in the solutions. So a company like Intel coming into this space is quite enormous.
Basically, it’s validating this space and it’s also saying look, we’re going to provide something that’s going to make this easier and enable new idea exploration. In 10 years, we’re going to have things that are going to blow our minds. You can all of a sudden do something 1000x faster than you could do before. I don’t even know what that’s going to enable.
Once you can do something 1000x faster, what more can you build? And that’s super-exciting for me. Things that enable new ways of thinking, that’s really why we did this.