课程: DJ Patil: Ask Me Anything

How do you feel about machine learning for business decisions?

课程: DJ Patil: Ask Me Anything

开始一个月试用

How do you feel about machine learning for business decisions?

“

(upbeat music) - Now for those that are actually in the capability to actually start using AI and machine learning, it's actually the thing I most tell people is don't go for the sexy problem. Go for the most boring problems. It's a boring problem actually where you get the disproportionate benefit, and as a concrete example of this, we put a team together for Secretary of Defense Ash Carter and said you have the full reign of The Pentagon. You can look across anywhere across the Department of Defense and figure out where AI and machine learning are going to have a big, disproportionate impact. Tell us where. So what did they come back with? Most people usually expect autonomous weapons and all these other kind of things. It was like how to move fuel, how to move batteries, more efficient movement of personnel, logistics. It's all logistics, and there's all these problems where the algorithms actually came out of the World War II era of logistic regression and then moved into linear programming, what are often referred to as convex whole-type problems where you think of it a big bowl, and you're trying to find the optimal state. That's the problem set where this has huge benefits. We're also seeing it in manufacturing where if you take the traditional manufacturing approach of how your engineer something and you use AI or machine learning, you may find other efficiencies. You kind of find these things around the edges, but just chasing after a generalized machine learning layer without actually having data first, people to enable this, and then the ability to actually use it or implement it. That's where you should start, and then you can get to these other things, but start with the boring things. Don't go for the sexy stuff.

内容